WO2024138482A1

WO2024138482A1 - Resource management method and corresponding apparatus

Info

Publication number: WO2024138482A1
Application number: PCT/CN2022/143113
Authority: WO
Inventors: 钱建民
Original assignee: 华为技术有限公司
Priority date: 2022-12-29
Filing date: 2022-12-29
Publication date: 2024-07-04

Abstract

The present application discloses a resource management method. The method comprises: receiving a resource request, wherein the resource request comprises the number of requested resource nodes; acquiring at least two sub-requests according to the resource request, wherein the sum of the number of resource nodes respectively corresponding to the at least two sub-requests is the number of resource nodes corresponding to the resource request; and allocating a corresponding number of resource nodes to each sub-request according to resource management information, wherein the resource management information comprises a hierarchical grouping relationship of a plurality of resource nodes, and the number of resource nodes corresponding to different sub-requests is matched with the number of resource nodes corresponding to one or more groups of the same hierarchy or different hierarchies in the hierarchical grouping relationship. According to the solution provided by the present application, resource nodes can be allocated to sub-requests, and the number of resource nodes corresponding to the sub-requests is small, so that the resource nodes can be flexibly allocated to the sub-requests, thereby reducing the probability of resource fragmentation, and increasing the resource utilization rate.

Description

A resource management method and corresponding device

Technical Field

The present application relates to the field of communication technology, and in particular to a resource management method and corresponding device.

Background technique

Currently, artificial intelligence (AI) or big data tasks usually require heterogeneous computing devices such as graphic processing units (GPUs) or neural network processing units (NPUs) to accelerate the running of tasks and shorten the running time of tasks. These tasks are usually run on the cloud. For cloud vendors, it is necessary to provide a heterogeneous resource management platform to distribute computing resources such as GPUs or NPUs to tenants to perform tasks in order to ensure the efficient operation of tenant tasks.

In a cluster, a server is usually equipped with multiple computing device cards of the same model such as GPUs or NPUs. These computing device cards, CPUs, and memory resources are requested and used by upper-layer applications. At the same time, these computing device cards are connected through some high-speed interconnect buses (such as high-speed serial computer expansion buses (peripheral component interconnect express, PCIe)) to ensure efficient transmission of data information between multiple computing device cards.

In a multi-tenancy scenario on the cloud, tenant applications have various resource demands. If we only consider the affinity between allocated resources and allocate resources of some tasks together to reduce the overhead on the data transmission path, a small number of computing devices on some servers will lose the opportunity to be allocated, resulting in a large amount of allocation fragmentation. Therefore, the current resource management method faces the problem of high resource fragmentation rate and low allocation rate.

Summary of the invention

The present application provides a resource management method for reducing resource fragmentation and improving resource allocation rate. The present application also provides a corresponding device, a computer-readable storage medium, and a computer program product.

A first aspect of the present application provides a resource management method, comprising: receiving a resource request, the resource request including the number of requested resource nodes; obtaining at least two sub-requests according to the resource request, the sum of the numbers of resource nodes corresponding to the at least two sub-requests being the number of resource nodes corresponding to the resource request; allocating a corresponding number of resource nodes to each sub-request according to resource management information, the resource management information including a hierarchical grouping relationship of multiple resource nodes; wherein the number of resource nodes corresponding to different sub-requests matches the number of resource nodes corresponding to one or more groups at the same level or different levels in the hierarchical grouping relationship.

The resource management method provided in the present application can be applied to scenarios such as cloud service systems, data centers, independent servers or network systems. A resource node can be a computing device card or a virtual machine (VM) in a cloud service system. A resource node can be a computing node in a data center, such as a server or a virtual machine. A resource node can be a computing device card in a server. A resource node can be a network device in a network system. Among them, the computing device card can be at least one of a central processing unit (CPU), a graphics processing unit (GPU) and a neural network processor (NPU). The network device can be at least one of a firewall, a router and a switch.

In this application, a resource request can be sent by a tenant to a cloud service system, a data center, or a network system through a terminal device. A tenant can be a resource user.

In this application, the number of requested resource nodes can be understood as the number of servers requested by the tenant, the number of computing device cards requested by the tenant, etc., such as: the tenant requests 5 computing device cards, or the tenant requests 8 computing device cards.

In the present application, resource management information may include one or more hierarchical grouping relationships, each hierarchical grouping relationship is used to represent the hierarchical grouping of multiple resource nodes in different clusters. In different application scenarios, the meaning of the cluster may be different. Taking the cloud service system as an example, if each server includes 8 computing device cards, then one server is a cluster, and 8 computing device cards are 8 resource nodes. The number of resource nodes corresponding to different sub-requests can be allocated through a hierarchical grouping relationship, or through different hierarchical grouping relationships, such as: the number of resource nodes corresponding to sub-request 1 is allocated from server 1 according to hierarchical grouping relationship 1. The computing device card, and the number of resource nodes corresponding to sub-request 2 is allocated from server 2 according to hierarchical grouping relationship 2.

In the present application, the hierarchical grouping relationship includes multiple levels, each level has one or more groups, different levels have different numbers of groups, different levels of groups correspond to different numbers of resource nodes, and the same level of groups correspond to the same number of resource nodes.

In the present application, the number of resource nodes corresponding to different sub-requests matches the number of resource nodes corresponding to one or more groups at the same level or different levels, which means: if the number of resource nodes corresponding to two sub-requests is the same, resource nodes can be allocated to the two sub-requests according to the resource nodes corresponding to different groups at the same level. If the number of resource nodes corresponding to the two sub-requests is different, resource nodes can be allocated to the two sub-requests from the resource nodes corresponding to groups at different levels.

In the present application, a method for obtaining at least two sub-requests can be to split a resource request into at least two sub-requests, and then allocate resource nodes to the sub-requests. Because the number of resource nodes corresponding to each sub-request is less than the number of resource nodes corresponding to the resource request, resource nodes can be flexibly allocated to the sub-requests, thereby reducing the probability of resource fragmentation and improving resource utilization.

In a possible implementation, the hierarchical grouping relationship includes multiple levels, wherein each level corresponds to multiple resource nodes, the number of groups at different levels is different, and the number of resource nodes corresponding to each group at the same level is the same.

In this possible implementation, take 8 resource nodes and 4 levels as an example, each level corresponds to these 8 resource nodes. The number of groups in each level is different, and the number of groups in these 4 levels can be 1, 2, 4 and 8 respectively. When a level has 1 group, this group corresponds to 8 resource nodes. When a level has 2 groups, each group corresponds to 4 resource nodes. When a level has 4 groups, each group corresponds to 2 resource nodes. When a level has 8 groups, each group corresponds to 1 resource node. It should be noted that the number of levels in the hierarchical grouping relationship and the number of resource nodes corresponding to the groups in each level are not limited to the example here, such as: when there are 6 resource nodes, there can be 3 levels, and the number of groups in these 3 levels can be 1, 2 and 6 respectively. When a level has one group, this group corresponds to 6 resource nodes. When a level has 2 groups, each group corresponds to 3 resource nodes. When there are 6 groups at one level, each group corresponds to 1 resource node. The present application provides different numbers of groups at different levels, and the number of resource nodes corresponding to groups at different levels is different, which can improve the speed of allocating resource nodes to sub-requests.

In a possible implementation, each grouping at a different level corresponds to an exponential power of 2 resource nodes.

In this possible implementation, the exponential power of 2 can be expressed as 2 ⁿ , where n is a natural number (0, 1, ...). It should be noted that the exponential power of 2 corresponding to each group at different levels is only one way, and each group at different levels can also correspond to an exponential power of 3 resource nodes, or each group at different levels corresponds to a positive integer multiple of m resource nodes, where m is a positive integer.

In a possible implementation, each sub-request corresponds to a power-of-two resource node.

In this possible implementation, when each grouping at different levels corresponds to an exponential power of 2 resource nodes, each sub-request corresponds to an exponential power of 2 resource nodes, so that the number of resource nodes corresponding to the sub-request can match the number of resource nodes corresponding to the grouping at a certain level, and the resource nodes corresponding to the entire group are allocated to one sub-request, which can reduce the probability of resource node fragmentation.

It should be noted that when the number of resource nodes corresponding to each group at different levels is configured in other ways, the sub-requests can be divided into the number of corresponding resource nodes according to the corresponding configuration method.

In a possible implementation, the information of each level in the hierarchical grouping relationship is recorded in the form of a linked list. The linked list of each level includes the node identifiers of the resource nodes contained in each group of the level. The node identifiers of the resource nodes in different groups are different.

In this possible implementation, the node identifier of the resource node can uniquely identify the resource node, for example, it can indicate a certain position of the resource node on a certain server, for example, computing device card 1 in server 1. The linked list can also include the level or/and the number of resource nodes corresponding to each group in the level. The linked list can include a header and a group, and the header can be used to record the level. The header and the group, as well as the groups after grouping, can be connected by pointers. The number of node identifiers in the group can be used to represent the number of resource nodes corresponding to the group. By recording the information of each level in the hierarchical grouping relationship in the form of a linked list, when allocating resource nodes to sub-requests, the matching level can be quickly determined from the linked list according to the number of resource nodes corresponding to the sub-requests, thereby increasing the speed of resource node allocation.

In one possible implementation, at least two sub-requests include a first sub-request, and a corresponding number of resource nodes are allocated to each sub-request based on resource management information, including: based on the number of resource nodes corresponding to the first sub-request, a resource node corresponding to a node identifier of a group is allocated to the first sub-request from a linked list of the target level; wherein the number of node identifiers in each group of the target level is the same as the number of resource nodes corresponding to the first sub-request.

In this possible implementation, when allocating resource nodes to sub-requests, the matching target level can be quickly determined according to the number of resource nodes corresponding to the sub-requests, thereby increasing the speed of resource node allocation.

In one possible implementation, at least two sub-requests include a first sub-request, and a corresponding number of resource nodes are allocated to each sub-request based on resource management information, including: based on the number of resource nodes corresponding to the first sub-request, resource nodes corresponding to node identifiers of multiple groups are allocated to the first sub-request from a linked list of one or more levels; wherein the sum of the numbers of resource nodes corresponding to the node identifiers in the multiple groups is the same as the number of resource nodes corresponding to the first sub-request, the multiple groups belong to the same level, or at least two of the multiple groups belong to different levels.

In this possible implementation, the resource nodes allocated to the first sub-request may also come from multiple groups, and the multiple groups may be located at the same level, may not all be located at the same level, or may all be located at different levels.

In one possible implementation, the hierarchical grouping relationship is recorded in the form of a binary tree, and each group at the end level of the binary tree includes a node identifier of a resource node, and the node identifiers in different groups are different; the grouping of the upper level includes the association relationship between the grouping of the upper level and the grouping of the end level, wherein the upper level is any level above the end level.

In this possible implementation, a binary tree is a tree-shaped data storage structure, and the nodes in the upper layer can be divided into two forks. The hierarchical grouping relationship in this application can be expressed in the form of a binary tree. Taking 8 resource nodes as an example, the topmost layer can have 1 group, and this 1 group corresponds to 8 resource nodes. The next layer divides the 1 group of the topmost layer into 2 groups, each of which corresponds to 4 resource nodes. In the next layer, each group of the previous layer is divided into 2 groups, and 4 groups are obtained, each of which corresponds to 2 resource nodes. In the next layer, each group of the previous layer is divided into 2 groups, and 8 groups are obtained, each of which corresponds to 1 resource node, and this layer becomes the last layer. The identifiers of the eight resource nodes can all be recorded in the grouping of the last level. Only the number of corresponding resource nodes needs to be recorded in the grouping of the upper level above the last level. When allocating resource nodes to sub-requests, the node identifier of the last level can be allocated to the sub-request based on the association between the upper level and the last level in the binary tree. By recording the hierarchical grouping relationship in the form of a binary tree, only the grouping of the last level needs to contain the node identifier of the resource node, and the other upper levels only need to store the number of resource nodes corresponding to the grouping, thereby reducing the amount of data storage.

In one possible implementation, at least two sub-requests include a second sub-request, and a corresponding number of resource nodes are allocated to each sub-request based on resource management information, including: based on the number of resource nodes corresponding to the second sub-request, resource nodes corresponding to the node identifier of the last level associated with the target group are allocated to the second sub-request, where the target group is a group of the target level of the binary tree; wherein the number of node identifiers of the last level associated with each group of the target level is the same as the number of resource nodes corresponding to the second sub-request.

In this possible implementation, when allocating resource nodes for sub-requests according to the binary tree, it is only necessary to determine the target level according to the number of resource nodes corresponding to the sub-request, and then allocate the node identifier of the last level associated with a group of the target level. It can be seen that the matching target level can be quickly determined from the binary tree according to the number of resource nodes corresponding to the sub-request, thereby improving the speed of resource node allocation.

In one possible implementation, at least two sub-requests include a second sub-request, and a corresponding number of resource nodes are allocated to each sub-request based on resource management information, including: based on the number of resource nodes corresponding to the second sub-request, resource nodes corresponding to the node identifiers of the last level associated with multiple target groups are allocated to the second sub-request, the multiple target groups belong to the same level, or at least two of the multiple groups belong to different levels; wherein the node identifiers of the last level associated with the multiple target groups are not repeated, and the sum of the number of node identifiers of the last level associated with the multiple target groups is the same as the number of resource nodes corresponding to the second sub-request.

In this possible implementation, the resource nodes allocated to the second sub-request may also come from multiple target groups, which may be located at the same level, or may not all be located at the same level, or may all be located at different levels. The node identifiers of the last level corresponding to these groups do not need to be repeated.

In a possible implementation, when at least two sub-requests are each corresponding to a number of resource nodes that satisfies an exponential power of 2, the resource request obtains the minimum number of sub-requests.

In this possible implementation, when there are multiple ways to obtain a resource request, the way to obtain the resource request with the least number of sub-requests is selected. This can reduce the number of sub-request allocations, improve allocation efficiency, and also improve the affinity between multiple resource nodes corresponding to the same sub-request.

In a possible implementation, the hierarchical grouping relationship is obtained by dividing a plurality of resource nodes by level, and when the number of resource nodes corresponding to a group at any level is greater than 2, at least two resource nodes corresponding to the group constitute a communication ring.

In this possible implementation, the communication ring refers to a ring formed by connecting at least two resource nodes end to end. That is, after data is sent on a resource node, it can traverse all other resource nodes in the ring and eventually return to the initial resource node. When the resource nodes are grouped in multiple levels, the present application no longer relies on the organization of the high-speed serial computer expansion bus (peripheral component interconnect express, PCIe) architecture, but is divided according to the communication ring, decoupled from the hardware interconnection architecture, compatible with the current mainstream architecture, and highly versatile.

In one possible implementation, the sum of the bandwidths of the communication rings corresponding to different groups at the same level is the same, and the bandwidth of the communication ring is the sum of the communication bandwidths between the resource nodes constituting the communication ring. For example, the communication ring includes 4 resource nodes, among which the bandwidth between resource node 1 and resource node 2 is 10M, the bandwidth between resource node 2 and resource node 3 is 10M, the bandwidth between resource node 3 and resource node 4 is 10M, and the bandwidth between resource node 4 and resource node 1 is 10M. Then the sum of the bandwidths of the communication ring = 10+10+10+10 = 40M.

In this possible implementation, the sum of the bandwidths of the communication rings formed when the resource nodes are divided is consistent, thereby avoiding the situation where the performance of the communication rings is inconsistent in the scenario of an asymmetric interconnection topology architecture.

In one possible implementation, the sum of the bandwidths of communication rings at the same level is the largest of at least two first bandwidth sums, wherein the first bandwidth sum is the sum of the bandwidths of the communication rings obtained when multiple resource nodes are grouped in different ways according to the number of resource nodes corresponding to the grouping at the same level.

In this possible implementation, if there are multiple ways to combine resource nodes corresponding to the same level into a communication ring, selecting a combination with the largest sum of bandwidths of the communication ring can improve the performance of the communication ring.

In a possible implementation, the resource node is located in a cloud service system, and the resource node is a computing device card or a virtual machine in the cloud service system.

In this possible implementation, the resources in the cloud service system usually provide services to tenants in the form of computing device cards or virtual machines. Therefore, in the cloud service system scenario, the resource node can be a computing device card or a virtual machine. If the resources in the cloud service system provide services to tenants in other forms, then the resource node can be other forms of devices, not limited to computing device cards or virtual machines.

In a possible implementation, the resource node is located in a data center, the resource node is a computing node of the data center, and the computing node is a server or a virtual machine.

In this possible implementation, the resources of the data center usually provide services to the outside in the form of servers or virtual machines. Therefore, in the data center scenario, the resource nodes can be servers or virtual machines. If the resources of the data center provide services to the outside in other forms, then the resource nodes can be other forms of devices, not limited to servers or virtual machines.

In a possible implementation, the resource node is located in a network system, and the resource node is a network device of the network system, and the network device includes at least one of a firewall, a router, and a switch.

In this possible implementation, the resources of the network system usually provide services to the outside in the form of firewalls, routers or switches. Therefore, in the network system scenario, the resource node can be at least one of the firewalls, routers and switches. If the resources of the network system provide services to the outside in other forms, then the resource node can be other forms of devices, not limited to at least one of the firewalls, routers and switches.

The second aspect of the present application provides a resource management method, including: obtaining a topological relationship between multiple resource nodes, the topological relationship representing the communication relationship between the multiple resource nodes; grouping the multiple resource nodes according to the topological relationship between the multiple resource nodes to obtain a hierarchical grouping relationship of the multiple resource nodes, the hierarchical grouping relationship is obtained by dividing the multiple resource nodes according to the hierarchy, the hierarchical grouping relationship includes multiple hierarchies, wherein each hierarchy corresponds to multiple resource nodes, the number of groups at different hierarchies is different, and each group at the same hierarchy corresponds to the same number of resource nodes.

In a possible implementation, when the number of resource nodes corresponding to a group at any level is greater than 2, at least two resource nodes corresponding to the group constitute a communication ring.

In this application, the topological relationship refers to the connection relationship between resource nodes, and each resource node can be connected through a bus. The communication ring can be understood by referring to the above introduction.

In this possible implementation, when performing multi-level grouping of resource nodes, it no longer relies on the organization of the high-speed serial computer expansion bus (peripheral component interconnect express, PCIe) architecture. Instead, it is divided according to the communication ring, decoupled from the hardware interconnection architecture, compatible with the current mainstream architecture, and highly versatile.

In a possible implementation, the sum of bandwidths of communication rings corresponding to different groups at the same level is the same, and the bandwidth of the communication ring is the sum of communication bandwidths between resource nodes constituting the communication ring.

In a possible implementation, each group at a different level contains a power-of-two resource node.

In one possible implementation, the hierarchical grouping relationship is recorded in the form of a binary tree, and each group at the end level of the binary tree includes a node identifier of a resource node, and the node identifiers in different groups are different; the grouping of the upper level includes the number of resource nodes corresponding to the grouping of the upper level, wherein the upper level is any level above the end level.

The features and intended effects of the second aspect and any possible implementation manner may be understood by referring to the first aspect or any possible implementation manner of the first aspect.

A third aspect of the present application provides a resource management device, including:

The receiving unit is used to receive a resource request, where the resource request includes the number of requested resource nodes.

The first processing unit is used to obtain at least two sub-requests according to the resource request received by the receiving unit, and the sum of the numbers of resource nodes corresponding to the at least two sub-requests is the number of resource nodes corresponding to the resource request.

The second processing unit is used to allocate a corresponding number of resource nodes to each sub-request obtained by the first processing unit according to resource management information, and the resource management information includes a hierarchical grouping relationship of multiple resource nodes; wherein the number of resource nodes corresponding to different sub-requests matches the number of resource nodes corresponding to one or more groups at the same level or different levels in the hierarchical grouping relationship.

In one possible implementation, the information of each level in the hierarchical grouping relationship is recorded in the form of a linked list. The linked list of each level includes the node identifiers of the resource nodes contained in each group of the level, and the node identifiers of the resource nodes in different groups are different.

In one possible implementation, at least two sub-requests include a first sub-request and a second processing unit, which is specifically used to allocate a resource node corresponding to a node identifier of a group to the first sub-request from a linked list of the target level according to the number of resource nodes corresponding to the first sub-request; wherein the number of node identifiers in each group of the target level is the same as the number of resource nodes corresponding to the first sub-request.

In one possible implementation, at least two sub-requests include a first sub-request and a second processing unit, which is specifically used to allocate resource nodes corresponding to node identifiers of multiple groups to the first sub-request from a linked list of one or more levels according to the number of resource nodes corresponding to the first sub-request; wherein the sum of the numbers of resource nodes corresponding to the node identifiers in the multiple groups is the same as the number of resource nodes corresponding to the first sub-request, the multiple groups belong to the same level, or at least two of the multiple groups belong to different levels.

In one possible implementation, at least two sub-requests include a second sub-request, and a second processing unit is specifically used to allocate a resource node corresponding to the node identifier of the last level associated with the target group to the second sub-request based on the number of resource nodes corresponding to the second sub-request, and the target group is a group of the target level of the binary tree; wherein the number of node identifiers of the last level associated with each group of the target level is the same as the number of resource nodes corresponding to the second sub-request.

In one possible implementation, at least two sub-requests include a second sub-request, and a second processing unit is specifically used to allocate resource nodes corresponding to node identifiers of the last level associated with multiple target groups to the second sub-request according to the number of resource nodes corresponding to the second sub-request, and the multiple target groups belong to the same level, or at least two of the multiple groups belong to different levels; wherein the node identifiers of the last level associated with the multiple target groups are not repeated, and the sum of the number of node identifiers of the last level associated with the multiple target groups is the same as the number of resource nodes corresponding to the second sub-request.

In one possible implementation, the resource node is located in the cloud service system, and the resource node is a computing device card or a virtual machine in the cloud service system.

A fourth aspect of the present application provides a resource management device, including:

The first processing unit is used to obtain a topological relationship between multiple resource nodes, where the topological relationship represents a communication relationship between the multiple resource nodes.

The second processing unit is used to group the multiple resource nodes according to the hierarchy based on the topological relationship between the multiple resource nodes to obtain a hierarchical grouping relationship of the multiple resource nodes. The hierarchical grouping relationship is obtained by dividing the multiple resource nodes according to the hierarchy. The hierarchical grouping relationship includes multiple levels, wherein each level corresponds to multiple resource nodes, the number of groups at different levels is different, and each group at the same level corresponds to the same number of resource nodes.

The fifth aspect of the present application provides a resource management device for executing the method in the first aspect or any possible implementation of the first aspect. Specifically, the resource management device includes a module or unit for executing the method in the first aspect or any possible implementation of the first aspect, such as: a receiving unit, a first processing unit, and a second processing unit.

In a sixth aspect of the present application, a resource management device is provided for executing the method in the second aspect or any possible implementation of the second aspect. Specifically, the resource management device includes a module or unit for executing the method in the second aspect or any possible implementation of the second aspect, such as a first processing unit and a second processing unit.

In a seventh aspect of the present application, a resource management device is provided. The resource management device may include at least one processor, a memory, and a communication interface. The processor is coupled to the memory and the communication interface. The memory is used to store instructions, the processor is used to execute the instructions, and the communication interface is used to communicate with other network elements under the control of the processor. When the instructions are executed by the processor, the processor executes the method in the first aspect or any possible implementation of the first aspect.

In an eighth aspect of the present application, a resource management device is provided. The resource management device may include at least one processor, a memory, and a communication interface. The processor is coupled to the memory and the communication interface. The memory is used to store instructions, the processor is used to execute the instructions, and the communication interface is used to communicate with other network elements under the control of the processor. When the instructions are executed by the processor, the processor executes the method in the second aspect or any possible implementation of the second aspect.

In the ninth aspect of the present application, a chip system is provided, which includes one or more interface circuits and one or more processors; the interface circuits and the processors are interconnected through lines; the interface circuits are used to receive signals from the memory of the resource management device and send signals to the processor, and the signals include computer instructions stored in the memory; when the processor executes the computer instructions, the resource management device executes the method in the aforementioned first aspect or any possible implementation of the first aspect.

The tenth aspect of the present application provides a chip system, which includes one or more interface circuits and one or more processors; the interface circuit and the processor are interconnected by lines; the interface circuit is used to receive signals from the memory of the resource management device and send signals to the processor, and the signals include computer instructions stored in the memory; when the processor executes the computer instructions, the resource management device executes the method in the aforementioned second aspect or any possible implementation of the second aspect.

The eleventh aspect of the present application provides a computer-readable storage medium on which a computer program or instruction is stored. When the computer program or instruction is executed on a computer device, the computer device executes the method in the aforementioned first aspect or any possible implementation of the first aspect.

A twelfth aspect of the present application provides a computer-readable storage medium on which a computer program or instruction is stored. When the computer program or instruction is executed on a computer device, the computer device executes the method in the aforementioned second aspect or any possible implementation of the second aspect.

The thirteenth aspect of the present application provides a computer device program product, which includes a computer device program code. When the computer device program code is executed on a computer device, the computer device executes the method in the aforementioned first aspect or any possible implementation of the first aspect.

A fourteenth aspect of the present application provides a computer device program product, which includes a computer device program code. When the computer device program code is executed on a computer device, the computer device executes the method in the aforementioned second aspect or any possible implementation of the second aspect.

A fifteenth aspect of the present application provides a resource management system, including a resource management device and multiple resource nodes, wherein the resource management device is used in the method in the aforementioned first aspect or any possible implementation of the first aspect.

A sixteenth aspect of the present application provides a resource management system, including a resource management device and multiple resource nodes, wherein the resource management device is used in the method in the aforementioned second aspect or any possible implementation of the second aspect.

Among them, the technical effects brought about by the second to sixteenth aspects or any possible implementation methods thereof can refer to the technical effects brought about by the first aspect or different possible implementation methods of the first aspect, and will not be repeated here.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

FIG1A is a schematic diagram of an architecture of a resource management system provided in an embodiment of the present application;

FIG1B is a schematic diagram of an architecture of a cloud service system provided in an embodiment of the present application;

FIG1C is a schematic diagram of an architecture of a data center provided in an embodiment of the present application;

FIG2 is a schematic diagram of a structure of a resource management device provided in an embodiment of the present application;

FIG3 is a schematic diagram of an embodiment of a resource management method provided in an embodiment of the present application;

FIG4 is a schematic diagram of an example of a hierarchical grouping process provided by an embodiment of the present application;

FIG5A is a schematic diagram showing an example of a hierarchical grouping relationship provided by an embodiment of the present application;

FIG5B is another exemplary schematic diagram of a hierarchical grouping relationship provided in an embodiment of the present application;

FIG5C is another exemplary schematic diagram of a hierarchical grouping relationship provided in an embodiment of the present application;

FIG6 is a schematic diagram of an embodiment of a resource management method provided in an embodiment of the present application;

FIG7A is a schematic diagram of a scenario example of a resource management method provided in an embodiment of the present application;

FIG7B is a schematic diagram of an updated linked list provided in an embodiment of the present application;

FIG8A is a schematic diagram of another scenario example of the resource management method provided in an embodiment of the present application;

FIG8B is a schematic diagram of an updated binary tree provided in an embodiment of the present application;

FIG9 is another schematic diagram of the structure of the resource management device provided in an embodiment of the present application;

FIG. 10 is another schematic diagram of the structure of the resource management device provided in an embodiment of the present application.

Detailed ways

The embodiment of the present application provides a resource management method for reducing resource fragmentation and improving resource allocation rate. The present application also provides a corresponding device, a computer-readable storage medium, and a computer program product, etc. The following are detailed descriptions.

The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of this application.

FIG. 1A is a schematic diagram of a structure of a resource management system provided in an embodiment of the present application.

As shown in FIG1A , the resource management system includes a resource management device and a plurality of resource nodes, and the resource management device can manage the plurality of resource nodes. The resource management device can group the plurality of resource nodes according to the topological relationship between the plurality of resource nodes according to the hierarchy to obtain a hierarchical grouping relationship of the plurality of resource nodes. The hierarchical grouping relationship is obtained by dividing the plurality of resource nodes according to the hierarchy, and the hierarchical grouping relationship includes a plurality of hierarchies, wherein each hierarchy corresponds to a plurality of resource nodes, and the number of groups at different hierarchies is different, and each group at the same hierarchy corresponds to the same number of resource nodes.

The resource management device saves the hierarchical grouping relationship of multiple resource nodes. After receiving a resource request, at least two sub-requests (such as sub-request 1 and sub-request 2 in FIG. 1A ) can be obtained according to the resource request, and then resource nodes are allocated to each sub-request according to the hierarchical grouping relationship of multiple resource nodes in the resource management information, such as: a resource node with a node identifier of X is allocated to sub-request 1, and at least two resource nodes with node identifiers of P to Q are allocated to sub-request 2.

The resource management device in FIG. 1A may be an independent device, or may be integrated on a control platform of a resource management system, or may be integrated on a certain resource node.

The functions of the resource management device can be implemented by software or hardware.

As an example of a software functional unit, the resource management device may include code running on a computing instance. The computing instance may include at least one of a physical host (computing device), a virtual machine, and a container. Furthermore, the computing instance may be one or more. For example, the resource management device may include code running on multiple hosts/virtual machines/containers. It should be noted that the multiple hosts/virtual machines/containers used to run the code may be distributed in the same region or in different regions. Furthermore, the multiple hosts/virtual machines/containers used to run the code may be distributed in the same availability zone (AZ) or in different AZs, each AZ including one data center or multiple data centers with close geographical locations. Generally, a region may include multiple AZs.

Similarly, multiple hosts/virtual machines/containers used to run the code can be distributed in the same virtual private cloud (VPC) or in multiple VPCs. Usually, a VPC is set up in one region. For cross-region communication between two VPCs in the same region and between VPCs in different regions, a communication gateway needs to be set up in each VPC to achieve interconnection between VPCs through the communication gateway.

As an example of a hardware functional unit, the resource management device may include at least one computing device, such as a server, etc. Alternatively, the resource management device may also be a device implemented using an application-specific integrated circuit (ASIC) or a programmable logic device (PLD). The PLD may be a complex programmable logical device (CPLD), a field-programmable gate array (FPGA), a generic array logic (GAL) or any combination thereof.

The multiple computing devices included in the resource management device can be distributed in the same area or in different areas. The multiple computing devices included in the resource management device can be distributed in the same AZ or in different AZs. Similarly, the multiple computing devices included in the resource management device can be distributed in the same VPC or in multiple VPCs. The multiple computing devices can be any combination of computing devices such as servers, ASICs, PLDs, CPLDs, FPGAs, and GALs.

The resource management system provided in the embodiment of the present application may be a cloud service system. In the cloud service system, as shown in FIG1B , the cloud service system includes a cloud platform and basic resources. The cloud platform includes a cloud platform manager, and the resource management device in FIG1A may be the cloud platform manager in FIG1B . The basic resources may include multiple servers, each of which may include multiple resource nodes.

The resource node in FIG1B may be a computing device card or a virtual machine (VM). The computing device card may be at least one of a central processing unit (CPU), a graphics processing unit (GPU), and a neural network processor (NPU).

Among them, when multiple resource nodes in each server are grouped according to the hierarchy, if there are at least two groups in the hierarchy, then the resource nodes corresponding to each group can form a communication ring. A communication ring refers to a ring formed by at least two resource nodes connected end to end. That is, after the data is sent on a resource node, it can traverse all other resource nodes in the ring and finally return to the initial resource node. When the present application performs multi-level grouping of resource nodes, it no longer relies on the organization of the high-speed serial computer expansion bus (peripheral component interconnect express, PCIe) architecture, but is divided according to the communication ring, decoupled from the hardware interconnection architecture, compatible with the current mainstream architecture, and has strong versatility.

In addition, the resource nodes in each server can obtain one or more hierarchical grouping relationships through multi-level grouping. For example, if a server includes 8 GPUs, 8 NPUs and 8 CPUs, the 8 GPUs can obtain a hierarchical grouping relationship, the 8 NPUs can obtain a hierarchical grouping relationship, and the 8 CPUs can obtain a hierarchical grouping relationship. If there is a connection between the GPU, NPU and CPU, a communication ring can be formed. The combination of GPU, NPU and CPU can also obtain a hierarchical grouping relationship, as long as the corresponding computing device cards in different hierarchical grouping relationships are not repeated.

The cloud platform scheduler maintains resource management information, which may include one or more hierarchical grouping relationships in each server. If the server in the cloud service system is understood as a cluster, each hierarchical grouping relationship is used to represent the hierarchical grouping of multiple resource nodes in different clusters, or the hierarchical grouping of multiple resource nodes in different subsets in the same cluster.

In a cloud service system, a resource request may be sent by a tenant to the cloud service system through a terminal device, and the tenant may be a resource user.

After receiving the resource request, the cloud platform manager can obtain at least two sub-requests according to the resource request, and then allocate a resource node to each sub-request according to the hierarchical grouping relationship.

The resource management system provided in the embodiment of the present application can be a data center. In the data center, as shown in FIG1C , the data center includes a data center management platform, an internal network of the data center, and multiple servers. Each server includes a hardware layer and a software layer. The hardware layer includes a memory, a network card, a processor, and a disk. The memory, the network card, the processor, and the disk are connected through a bus. The hardware layer provides the hardware resources required for the operation of the virtual machines in the software layer. The software layer includes a host operating system and multiple virtual machines. The host operating system may include a data center management platform client, which can interact with the data center management platform.

Virtualization technology mainly consists of computing virtualization and input/output (I/O) virtualization. It shares a physical server with multiple tenants at the granularity of virtual machines, allowing tenants to use physical resources conveniently and flexibly under the premise of secure isolation, and can greatly improve the utilization rate of physical resources.

Computing virtualization is the process of providing computing resources such as the server's processor and memory to virtual instances. For example, virtual instances are virtual machines. In other scenarios, virtual instances are containers and bare metal servers.

In Fig. 1C, each server obtains multiple virtual machines through virtualization technology, and each virtual machine can be understood as a resource node. The resource management device in Fig. 1A can be the data center management platform in Fig. 1C.

Among them, virtual machines can also be called cloud servers (Elastic Compute Service, ECS) and elastic instances (different cloud service providers have different names).

Among them, when multiple virtual machines in each server can be grouped according to levels, a hierarchical grouping relationship is obtained for use when subsequent tenants request resources.

The data center management platform can provide an access interface (such as an interface or an application programming interface (API)). The tenant can operate the client remote access access interface to register an account and password in the data center management platform and log in to the data center management platform. After the data center management platform successfully authenticates the account and password, the tenant can send a resource request to the data center management platform through the client, and then obtain at least two sub-requests based on the resource request, and allocate virtual machines to the sub-requests based on the hierarchical grouping relationship stored in the data center management platform. For example, the identifier of virtual machine 1 is allocated to sub-request 1, and the identifier of virtual machine 2 is allocated to sub-request 2.

In addition, the resource management system provided in the embodiment of the present application can be a network system. In the network system, the resource nodes in the network system are network devices, and the network devices can be at least one of a firewall, a router and a switch.

FIG2 is a possible logical structure diagram of a resource management device provided in an embodiment of the present application. As shown in FIG2 , the resource management device 20 provided in an embodiment of the present application includes: a processor 201, a communication interface 202, a memory 203 and a bus 204. The processor 201, the communication interface 202 and the memory 203 are interconnected via the bus 204. In an embodiment of the present application, the processor 201 is used to control and manage the actions of the resource management device 20, for example, the processor 201 is used to execute the process of determining audio advertisements. The communication interface 202 is used to support the resource management device 20 to communicate, for example: the communication interface 202 can execute the steps of receiving advertisement requests and sending audio advertisements. The memory 203 is used to store the program code and data of the resource management device 20.

Among them, the processor 201 can be a central processing unit, a general processor, a digital signal processor, an application-specific integrated circuit, a field programmable gate array or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It can implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of this application. The processor can also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a digital signal processor and a microprocessor, and the like. The bus 204 can be a peripheral component interconnect standard (Peripheral Component Interconnect, PCI) bus or an extended industry standard architecture (Extended Industry Standard Architecture, EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in Figure 2, but it does not mean that there is only one bus or one type of bus.

The resource management method provided by the embodiment of the present application is described below. The content involved in the method being executed by the resource management device can be executed by the resource management device, or by a component of the resource management device (such as a processor, chip, or chip system, etc.).

In the embodiment of the present application, the resource management device can manage the resource nodes in the resource management system. One resource management scheme is: multi-level grouping of multiple resource nodes; another resource management scheme is: allocating resource nodes to resource requests. The following are introduced respectively:

In conjunction with the resource management system introduced in FIG. 1A to FIG. 1C , the resource management solution for multi-level grouping of resource nodes can be understood by referring to FIG. 3 .

As shown in FIG3 , an embodiment of the resource management method provided in the embodiment of the present application includes:

301. The resource management device obtains a topological relationship between multiple resource nodes, where the topological relationship represents a communication relationship between the multiple resource nodes.

302. The resource management device groups the multiple resource nodes according to the topological relationship between the multiple resource nodes to obtain a hierarchical grouping relationship of the multiple resource nodes.

The hierarchical grouping relationship is obtained by dividing multiple resource nodes according to the hierarchy. The hierarchical grouping relationship includes multiple levels, wherein each level corresponds to multiple resource nodes, different levels have different numbers of groups, and different levels of groups correspond to different numbers of resource nodes, while each group at the same level corresponds to the same number of resource nodes.

The resource management device in the present application can obtain the topological relationship of the resource nodes before dividing the resource nodes. For example, taking the cloud service system shown in FIG. 1B as an example, the interconnection topology between the computing device cards on a single server can be obtained through the topology acquisition interface provided by the hardware driver of the computing device card. Then, the resource management device can group multiple resource nodes in multiple levels according to the topological relationship. Multi-level grouping means that multiple resource nodes are grouped at each level, but the number of groups is different. When grouping at each level, the dichotomy method can be used as the basis for the previous level until each group corresponds to a resource node, and the division of the resource nodes is ended.

Taking 8 resource nodes as an example, in the sequential grouping from the upper layer to the lower layer, the top layer can have 1 group corresponding to these 8 resource nodes. The next layer can divide the 8 resource nodes corresponding to the upper layer group into 2 groups, each of which corresponds to 4 resource nodes. The next layer can divide the 4 resource nodes corresponding to each of the 2 upper layer groups into 2 groups, obtaining 4 groups, each of which corresponds to 2 resource nodes. The next layer can divide the 4 groups corresponding to the 2 resource nodes in the upper layer into 2 groups, obtaining 8 groups, each of which corresponds to 1 resource node. The grouping process can be understood by referring to Figure 4.

As shown in Figure 4, take a server with 8 resource nodes as an example, namely resource node 0, resource node 1, ..., resource node 7. If resource node 0, resource node 1, resource node 2 and resource node 3 can form a communication ring, resource node 4, resource node 5, resource node 6 and resource node 7 can form a communication ring, then the 8 resource nodes can be divided into 2 groups, each group corresponding to 4 resource nodes. Then divide each group corresponding to the 4 resource nodes, if resource node 0 and resource node 1 can form a communication ring, resource node 2 and resource node 3 can form a communication ring, then resource node 0 and resource node 1 are divided into one group, and resource node 2 and resource node 3 are divided into one group. Similarly, the group corresponding to resource node 4, resource node 5, resource node 6 and resource node 7 can also be divided in this way, resource node 4 and resource node 5 are divided into one group, and resource node 6 and resource node 7 are divided into one group. Then, divide again for the 4 groups corresponding to the 2 resource nodes, and divide each resource node into a group separately, and the division ends.

After the 8 resource nodes listed above are divided, the resulting hierarchical grouping relationship can include 4 levels, and the number of groups in these 4 levels can be 1, 2, 4, and 8 respectively. When a level has 1 group, this group corresponds to 8 resource nodes. When a level has 2 groups, each group corresponds to 4 resource nodes. When a level has 4 groups, each group corresponds to 2 resource nodes.

It should be noted that when dividing resource nodes, it is not necessary to divide them into 8 resource nodes, and other numbers of resource nodes can also be divided. Therefore, the number of levels in the hierarchical grouping relationship obtained after the division, and the number of resource nodes corresponding to the groups in each level are not limited to the 8 resource nodes in the above example. For example, when there are 6 resource nodes, there can be 3 levels, and the number of groups in these 3 levels can be 1, 2 and 6 respectively. When a level has one group, this group corresponds to 6 resource nodes. When a level has 2 groups, each group corresponds to 3 resource nodes. When a level has 6 groups, each group corresponds to 1 resource node.

The examples listed above all use binary search to group multiple resource nodes. If the number of resource nodes is an odd number and cannot be directly divided into two equal parts, most of the resource nodes can be divided into one group through a single division so that the number of resource nodes in the group is an even number or an exponential power of 2. The subsequent division of the group can be carried out in the form of binary search. For example, a resource cluster includes 17 resource nodes. During the first division, 16 of them can be divided into one group, and then the 16 resource nodes in the group can be subsequently divided by binary search.

The multi-level grouping process illustrated in the above FIG4 is divided on the premise that at least two resource nodes in each group form a communication ring. In this way, when performing multi-level grouping of resource nodes, it is no longer dependent on the organization of the high-speed serial computer expansion bus (peripheral component interconnect express, PCIe) architecture. Instead, it is divided according to the communication ring, decoupled from the hardware interconnection architecture, compatible with the current mainstream architecture, and highly versatile.

It should be noted that in an embodiment of the present application, even if the resource nodes in a group cannot form a communication ring, the above-mentioned multi-level grouping method can be used to obtain a hierarchical grouping relationship. Managing the resource nodes according to the hierarchical grouping relationship can improve the efficiency of resource node management, which is conducive to quickly allocating resource nodes to resource requests after receiving resource requests.

When grouping according to the conditions for forming a communication ring, the sum of the bandwidths of the communication rings corresponding to different groups at the same level is the same, and the bandwidth of the communication ring is the sum of the communication bandwidths between the resource nodes that constitute the communication ring. For example: the communication ring includes 4 resource nodes, among which the bandwidth between resource node 1 and resource node 2 is 10M, the bandwidth between resource node 2 and resource node 3 is 10M, the bandwidth between resource node 3 and resource node 4 is 10M, and the bandwidth between resource node 4 and resource node 1 is 10M, then the sum of the bandwidth of the communication ring = 10+10+10+10 = 40M. The sum of the bandwidths of the communication rings formed when the resource nodes are divided is consistent, which avoids the situation where the performance of the communication ring is inconsistent in the scenario of an asymmetric interconnection topology architecture.

Optionally, in an embodiment of the present application, the sum of the bandwidths of the communication rings at the same level is the largest of at least two first bandwidth sums, wherein the first bandwidth sum is the sum of the bandwidths of the communication rings obtained when multiple resource nodes are grouped in accordance with the number of resource nodes corresponding to the same level and are combined in different ways. In other words, if there are multiple ways to combine the resource nodes corresponding to the same level into a communication ring, select the combination method with the largest sum of the bandwidths of the communication ring. For example: if resource node 0, resource node 1, resource node 2, and resource node 3 can form a communication ring, in this combination method, the bandwidth of the communication ring is 40M, resource node 0, resource node 1, resource node 2, and resource node 5 can also form a communication ring, in this combination method, the bandwidth of the communication ring is 35M, then select a combination method in which resource node 0, resource node 1, resource node 2, and resource node 3 are divided into one group.

Optionally, the hierarchical grouping relationship provided in the embodiment of the present application may be recorded in the form of a linked list or in the form of a binary tree, which are respectively introduced below.

1. The hierarchical grouping relationship is recorded in the form of a linked list.

The hierarchical grouping relationship of the eight resource nodes shown in FIG. 4 can be understood by referring to FIG. 5A .

As shown in FIG5A, according to the order of resource nodes from few to many, the 8 resource nodes in FIG4 can be divided into 4 levels, and these 4 levels can be recorded in order from bottom to top as level 1, level 2, level 3 and level 4. Of course, the order of the levels can also be recorded in reverse, and FIG5A is only used as an example to illustrate the order from bottom to top. Among them, each group of level 1 corresponds to 1 resource node, each group of level 2 corresponds to 2 resource nodes, each group of level 3 corresponds to 4 resource nodes, and 1 group of level 4 corresponds to 8 resource nodes.

The hierarchical grouping relationship shown in FIG5A is recorded in the form of a linked list, and the linked list of each level includes the node identifiers of the resource nodes contained in each group of the level, and the node identifiers of the resource nodes in different groups are different. The node identifier of the resource node can uniquely identify the resource node, for example, it can indicate a certain position of the resource node on a certain server, for example, computing device card 1 in server 1.

The linked list may also include the level or/and the number of resource nodes corresponding to each group in the level. The linked list may include a header and a group. The header may be used to record the level. The header and the group, as well as the groups after grouping, may be connected by pointers. The number of node identifiers in a group may be used to represent the number of resource nodes corresponding to the group. Of course, the number of resource nodes corresponding to each group may also be recorded in the header.

2. The hierarchical grouping relationship is recorded in the form of a binary tree.

The hierarchical grouping relationship in the embodiment of the present application can also be recorded in the form of a binary tree.

A binary tree is a tree-shaped data storage structure, and the nodes in the upper layer can be divided into two forks. The hierarchical grouping relationship in this application can be expressed in the form of a binary tree. Taking 8 resource nodes as an example, the topmost layer can have 1 group, and this 1 group corresponds to 8 resource nodes. The next layer divides the 1 group of the topmost layer into 2 groups, each of which corresponds to 4 resource nodes. In the next layer, each group of the previous layer is divided into 2 groups, and 4 groups are obtained, each of which corresponds to 2 resource nodes. In the next layer, each group of the previous layer is divided into 2 groups, and 8 groups are obtained, each of which corresponds to 1 resource node, and this layer becomes the last layer.

In the embodiment of the present application, the identifiers of the eight resource nodes may be recorded in the group at the end level, or may be recorded in the group at each level. Please refer to FIG. 5B and FIG. 5C for understanding.

1. The identification of the resource node is recorded in the group at the end level.

The node identifiers of the eight resource nodes can all be recorded in the grouping of the last level. As shown in FIG5B , each grouping of the last level (level 1) of the binary tree includes a node identifier of a resource node, which are node identifiers 0, ... node identifiers 7, respectively, and the node identifiers in different groups are different; the grouping of the upper levels (level 2, level 3, level 4) includes the association relationship between the grouping of the upper levels and the grouping of the last level (the association relationship can be represented by the arrow relationship between the upper level and the lower level in FIG5B ), wherein the upper level is any level above the last level. The number of corresponding resource nodes can be recorded in the grouping of the upper levels above the last level. As shown in FIG5B , the number of corresponding resource nodes 2 is recorded in the grouping of level 2, the number of corresponding resource nodes 4 is recorded in the grouping of level 3, and the number of corresponding resource nodes 8 is recorded in the grouping of level 4.

2. The identification of the resource node is recorded in each level of grouping.

The node identifiers of the eight resource nodes can be recorded in the groups of each level. As shown in FIG5C , each group of the last level (level 1) of the binary tree includes a node identifier of a resource node, which are node identifiers 0, ..., and node identifiers 7, respectively. The node identifiers in different groups are different. The four groups of level 2 respectively

record node identifiers

0, 1,

node identifiers

2, 3,

node identifiers

4, 5, and

node identifiers

6, 7. The two groups of level 3 respectively

record node identifiers

0, 1, 2, 3, and

node identifiers

4, 5, 6, 7. The one group of level 4

records node identifiers

0, 1, 2, 3, 4, 5, 6, 7.

The above describes a resource management solution for dividing resource nodes in a resource cluster. The following describes a resource management solution for allocating resource nodes to tenants' resource requests.

In conjunction with the resource management system described in FIG. 1A to FIG. 1C , the resource management solution for allocating resource nodes to resource requests may be understood by referring to FIG. 6 .

As shown in FIG6 , an embodiment of the resource management method provided in the embodiment of the present application includes:

601. The resource management device receives a resource request, where the resource request includes the number of requested resource nodes.

The number of resource nodes requested can be understood as the number of servers requested by the tenant, the number of computing device cards requested by the tenant, etc. For example, the tenant requests 5 computing device cards, or the tenant requests 8 computing device cards.

602. The resource management device obtains at least two sub-requests according to the resource request, and the sum of the numbers of resource nodes corresponding to the at least two sub-requests is the number of resource nodes corresponding to the resource request.

For example, the number of resource nodes corresponding to the resource request is 5. Sub-request 1 and sub-request 2 are obtained according to the resource request. Sub-request 1 corresponds to 1 resource node, and sub-request 2 corresponds to 4 resource nodes. 1+4=5.

603. The resource management device allocates a corresponding number of resource nodes to each sub-request based on the resource management information, and the resource management information includes a hierarchical grouping relationship of multiple resource nodes; wherein the number of resource nodes corresponding to different sub-requests matches the number of resource nodes corresponding to one or more groups at the same level or different levels in the hierarchical grouping relationship.

Resource management information may include one or more hierarchical grouping relationships, each hierarchical grouping relationship is used to represent the hierarchical grouping of multiple resource nodes in different clusters. In different application scenarios, the meaning of cluster may be different. Taking the cloud service system as an example, if each server includes 8 computing device cards, then one server is a cluster, and 8 computing device cards are 8 resource nodes. The number of resource nodes corresponding to different sub-requests can be allocated through a hierarchical grouping relationship, or through different hierarchical grouping relationships, such as: the number of resource nodes corresponding to sub-request 1 is allocated from server 1 according to hierarchical grouping relationship 1. The number of resource nodes corresponding to sub-request 2 is allocated from server 2 according to hierarchical grouping relationship 2.

In this application, the hierarchical grouping relationship can be understood by referring to the previous introduction to the resource node grouping part.

In the present application, the number of resource nodes corresponding to different sub-requests matches the number of resource nodes corresponding to one or more groups at the same level or different levels, which means: if the number of resource nodes corresponding to two sub-requests is the same, resource nodes can be allocated to the two sub-requests according to the resource nodes corresponding to different groups at the same level. For example, if the number of resource nodes corresponding to two sub-requests is 1, resource nodes can be allocated to the two sub-requests from two different groups at the same level, such as: allocating resource node 0 from the first group to sub-request 1, and allocating resource node 1 from the second group to sub-request 2.

If the number of resource nodes corresponding to two sub-requests is different, resource nodes can be allocated to the two sub-requests from resource nodes corresponding to groups at different levels. For example, if the number of resource nodes corresponding to sub-request 1 is 1 and the number of resource nodes corresponding to sub-request 2 is 4, resource node 0 can be allocated to sub-request 1 from the first group at level 1, and

resource nodes

4, 5, 6, and 7 from the second group at level 3 can be allocated to sub-request 2.

The solution provided in the embodiment of the present application can obtain at least two sub-requests based on the resource request, and then allocate resource nodes to the sub-requests. Because the number of resource nodes corresponding to each sub-request is less than the number of resource nodes corresponding to the resource request, resource nodes can be flexibly allocated to the sub-requests, thereby reducing the probability of resource fragmentation and improving resource utilization.

Optionally, in the embodiment of the present application, in the hierarchical grouping relationship, each grouping at a different level corresponds to a number of resource nodes which is an exponential power of 2. The exponential power of 2 can be expressed as 2 ⁿ , where n is a natural number (0, 1, ...).

It should be noted that each group at different levels corresponds to an exponential power of 2 resource nodes, which is only one way. Each group at different levels can also correspond to an exponential power of 3 resource nodes, or each group at different levels can correspond to a positive integer multiple of m resource nodes, where m is a positive integer.

Optionally, when each grouping at different levels corresponds to an exponential power of 2 resource nodes, each sub-request corresponds to an exponential power of 2 resource nodes, so that the number of resource nodes corresponding to the sub-request can match the number of resource nodes corresponding to the grouping at a certain level, and the resource nodes corresponding to the entire group are allocated to one sub-request, which can reduce the probability of resource node fragmentation.

It should be noted that, when the number of resource nodes corresponding to each group at different levels is configured in other ways, the sub-requests may be divided into the number of corresponding resource nodes according to the corresponding configuration methods.

Optionally, as can be seen from the foregoing description, the information of each level in the hierarchical grouping relationship is recorded in the form of a linked list, for example, the linked list of each level shown in FIG. 5A . For the introduction of the linked list, please refer to the foregoing content for understanding.

When the information of each level in the hierarchical grouping relationship is recorded in the form of a linked list, at least two sub-requests include the first sub-request, and the above step 603 may include: according to the number of resource nodes corresponding to the first sub-request, from the linked list of the target level, assigning a resource node corresponding to the node identifier of a group to the first sub-request; wherein the number of node identifiers in each group of the target level is the same as the number of resource nodes corresponding to the first sub-request. In this way, when assigning resource nodes to the sub-requests, the matching target level can be quickly determined according to the number of resource nodes corresponding to the sub-requests, thereby improving the speed of resource node assignment.

This situation can be understood by referring to FIG. 7A. As shown in FIG. 7A, the resource management system includes resource cluster 1 and resource cluster 2. The topological relationship of the resource nodes in resource cluster 1 and resource cluster 2 is different. Resource cluster 1 and resource cluster 2 can be understood as heterogeneous clusters. Resource cluster 1 and resource cluster 2 can be two servers or two units. There are multiple resource nodes in resource cluster 1. The node identification of the resource node can start from resource node 100 and be numbered as resource node 101, resource node 102, ..., and similarly, in resource cluster 2, the identity of the resource node can start from resource node 200 and be numbered as resource node 201, resource node 202, .... The node identification of the resource node is not illustrated in FIG. 7A. It can be understood by referring to the mark of FIG. 5A, except that the value of the node identification is different. Usually, in the scenario shown in FIG. 7A, the number of resource nodes in each resource cluster is usually a multiple of 8.

The resource management device determines linked list 1 of resource cluster 1 according to the topological relationship of resource cluster 1 , and determines linked list 2 of resource cluster 2 according to the topological relationship of resource cluster 2 .

There are four levels in the linked list 1, which are marked as level 1, level 2, level 3 and level 4 in FIG. 7A, where the level indicates the level.

Link list 2 also has four layers, which are marked as layer 1, layer 2, layer 3 and layer 4 in FIG. 7A , respectively, where the layers represent the hierarchy.

In linked list 1 or linked list 2, every 8 resource nodes can obtain a hierarchical grouping relationship such as that shown in FIG. 5A. Both resource cluster 1 and resource cluster 2 include topologies of multiple groups of 8 resource nodes. In linked list 1 or linked list 2, the hierarchical grouping relationship obtained for every 8 resource nodes can be connected by pointers after each level.

The resource management device receives resource request 1 and resource request 2. The number of resource nodes requested by resource request 1 is 5, indicating that the tenant wants to request 5 resource nodes. The number of resource nodes requested by resource request 2 is 6, indicating that the tenant wants to request 6 resource nodes.

The resource management device may obtain sub-request 1 and sub-request 2 according to resource request 1, wherein sub-request 1 is used to request 4 resource nodes, and sub-request 2 is used to request 1 resource node. The resource management device may obtain sub-request 3 and sub-request 4 according to resource request 2, wherein sub-request 3 is used to request 4 resource nodes, and sub-request 4 is used to request 2 resource nodes.

The resource management device can allocate 4 resource nodes to sub-request 1 from a group in layer 3 according to linked list 1, such as: resource node 100, resource node 101, resource node 102 and resource node 103 in resource cluster 1, and allocate 1 resource node to sub-request 2 from a group in layer 1, such as: resource node 104 in resource cluster 1.

The resource management device can allocate 4 resource nodes to sub-request 3 from a group of layer 3 according to linked list 2, such as: resource node 200, resource node 201, resource node 202 and resource node 203 in resource cluster 2, and allocate 2 resource nodes to sub-request 4 from a group of layer 2, such as: resource node 204 and resource node 205 in resource cluster 2.

The above example introduces the case of allocating resource nodes to sub-requests from the same group. In fact, resource nodes can also be allocated to sub-requests from multiple groups. The process can be: when at least two sub-requests include a first sub-request, according to the number of resource nodes corresponding to the first sub-request, resource nodes corresponding to the node identifiers of multiple groups are allocated to the first sub-request from one or more levels of linked lists; wherein the sum of the number of resource nodes corresponding to the node identifiers in the multiple groups is the same as the number of resource nodes corresponding to the first sub-request, the multiple groups belong to the same level, or at least two of the multiple groups belong to different levels.

Exemplarily, the resource management device may allocate 4 resource nodes to sub-request 1 from two groups in layer 2 according to linked list 1, such as resource nodes 100, resource nodes 101, resource nodes 102, and resource nodes 103 in resource cluster 1, and allocate 1 resource node to sub-request 2 from one group in layer 1, such as resource node 104 in resource cluster 1. Alternatively, the resource management device may allocate 4 resource nodes to sub-request 1 from four groups in layer 1 according to linked list 1, such as resource nodes 100, resource nodes 101, resource nodes 102, and resource nodes 103 in resource cluster 1. Alternatively, the resource management device may allocate 2 resource nodes to sub-request 1 from one group in layer 2 according to linked list 1, and allocate 2 resource nodes to sub-request 1 from two groups in layer 1. Alternatively, the resource management device may allocate 2 resource nodes to sub-request 1 from one group in layer 2 according to linked list 1, and then allocate 2 resource nodes to sub-request 1 from one group in layer 2 according to linked list 2. There may be many specific allocations, which are not limited in this application.

It should be noted that after allocating resource nodes, the resource management device can update the structures of linked

lists

1 and 2. Taking linked list 1 as an example, the updated structure of linked list 1 can be understood by referring to FIG. 7B. As shown in FIG. 7B, the first five of the original first eight resource node positions in layer 1 are eliminated, and the last three can continue to be allocated. Because five of the original first eight resource node positions in layer 2 are allocated, the first three combinations of two resource nodes are eliminated, leaving only one group representing a combination of two resource nodes.

Layers

3 and 4 are eliminated because five of the first eight resource node positions are allocated. If there is a sub-request for requesting 1 or 2 resource nodes, the resource management device can allocate resource node 105, resource node 106 and resource node 107 for the sub-request.

In an embodiment of the present application, the resource management device can allocate resource nodes to the sub-requests according to the number of resource nodes corresponding to the sub-requests, from the level that meets the number requirement and corresponds to the least number of resource nodes. In this way, the allocation speed of resource nodes can be improved.

Optionally, when the hierarchical grouping relationship can be recorded in the form of a binary tree, the above step 603 may include: according to the number of resource nodes corresponding to the second sub-request, allocating to the second sub-request the resource node corresponding to the node identifier of the last level associated with the target group, the target group being a group of the target level of the binary tree; wherein the number of node identifiers of the last level associated with each group of the target level is the same as the number of resource nodes corresponding to the second sub-request.

For the introduction of binary trees, please refer to the previous introduction for understanding.

Taking the binary tree form illustrated in FIG5B as an example, the resource node allocation process in this case can be understood by referring to FIG8A. As shown in FIG8A, the resource management system includes resource cluster 1 and resource cluster 2. The topological relationship of the resource nodes in resource cluster 1 and resource cluster 2 is different. Resource cluster 1 and resource cluster 2 can be understood as heterogeneous clusters. Resource cluster 1 and resource cluster 2 can be two servers or two units. There are 8 resource nodes in resource cluster 1. The node identification of the resource node can start from resource node 10 and be numbered as resource node 11, resource node 12, ..., resource node 17 in sequence. Similarly, in resource cluster 2, the node identification of the resource node can start from resource node 20 and be numbered as resource node 21, resource node 22, ..., resource node 27 in sequence.

The resource management device determines the binary tree 1 of the resource cluster 1 according to the topological relationship of the resource cluster 1 , and determines the binary tree 2 of the resource cluster 2 according to the topological relationship of the resource cluster 2 .

The resource management device receives a resource request, and the number of resource nodes requested by the resource request is 10, indicating that the tenant wants to request 10 resource nodes.

The resource management device may obtain sub-request 1 and sub-request 2 according to the resource request, wherein sub-request 1 is used to request 8 resource nodes, and sub-request 2 is used to request 2 resource nodes.

The resource management device can allocate 8 resource nodes to sub-request 1 from level 4 based on binary tree 1, such as resource node 10, resource node 11,... resource node 17 in resource cluster 1, and allocate 2 resource nodes to sub-request 2 from level 2 of binary tree 2, such as resource node 20 and resource node 21 in resource cluster 2.

The above example introduces the case of allocating resource nodes to sub-requests from the same target group. In fact, resource nodes can also be allocated to sub-requests from multiple target groups. The process can be: when at least two sub-requests include a first sub-request, resource nodes corresponding to node identifiers of the most terminal levels associated with multiple target groups are allocated to the second sub-request based on the number of resource nodes corresponding to the second sub-request. The multiple target groups belong to the same level, or at least two of the multiple groups belong to different levels; wherein the node identifiers of the most terminal levels associated with the multiple target groups are not repeated, and the sum of the number of node identifiers of the most terminal levels associated with the multiple target groups is the same as the number of resource nodes corresponding to the second sub-request.

Exemplarily, the resource management device may obtain sub-request 1 according to the resource request, wherein sub-request 1 is used to request 8 resource nodes.

The resource management device can allocate a total of 8 resource nodes to sub-request 1 from two groups of level 3 that each correspond to 4 resource nodes, such as

resource nodes

10 and 11 in resource cluster 1, ... resource nodes 17 such as

resource nodes

20 and 21 in resource cluster 2, according to binary tree 1. It is also possible to allocate 4 resource nodes to sub-request 1 from a group of level 3, and allocate another 4 resource nodes from two groups of level 2 that each correspond to two resource nodes. Of course, other allocation methods can also be used, and some resource nodes can also be allocated according to binary tree 1, and some resource nodes can be allocated according to binary tree 2. There can be many specific allocations, and this application does not limit this.

It should be noted that after allocating resource nodes, the resource management device will update the structure of binary tree 1 and binary tree 2. The updated binary tree 1 and binary tree 2 can be understood by referring to FIG8B. As shown in FIG8B, the eight positions representing resource nodes in binary tree 1 are set to gray, indicating that they have been allocated, and the first two positions representing resource nodes in binary tree 2 are set to gray, indicating that they have been allocated. Of course, the present application is not limited to setting gray, and it can also be indicated by other means that the resource node at the position has been allocated.

In an embodiment of the present application, the resource management device can allocate resource nodes to the sub-requests according to the number of resource nodes corresponding to the sub-requests, from the tree branch at the level that meets the number requirement and corresponds to the least number of resource nodes. In this way, the allocation speed of resource nodes can be improved.

Optionally, the at least two sub-requests are the minimum number of sub-requests obtained by the resource request when the number of resource nodes corresponding to each sub-request satisfies an exponential power of 2.

Taking the example of splitting a resource request to obtain at least two sub-requests, when there are multiple ways to split a resource request, the splitting method with the least number of sub-requests is selected. For example, the resource request in FIG8A above requests 10 resource nodes, and there are multiple ways to split in the form of exponential power of 2, for example: it can be split into 8 and 2, or it can be split into 4, 4 and 2. The splitting method of 8 and 2 is preferred. In this way, the number of sub-request allocations can be reduced, the allocation efficiency can be improved, and the affinity between multiple resource nodes corresponding to the same sub-request can be improved. Of course, if the remaining allocatable resource nodes in the resource management system cannot meet the allocation of the splitting method of 8 and 2, the splitting method of 4, 4 and 2 can also be used to allocate resource nodes to the tenant from different resource clusters.

The resource management method is introduced above. The resource management device in the embodiment of the present application is introduced below with reference to the accompanying drawings.

As shown in FIG9 , a structure of a resource management device 90 provided in an embodiment of the present application includes:

The first processing unit 901 is used to obtain a topological relationship between multiple resource nodes, where the topological relationship represents a communication relationship between multiple resource nodes. The first processing unit 901 can execute step 301 in the above method embodiment.

The second processing unit 902 is used to group the multiple resource nodes according to the hierarchical relationship between the multiple resource nodes to obtain a hierarchical grouping relationship of the multiple resource nodes, wherein the hierarchical grouping relationship is obtained by dividing the multiple resource nodes according to the hierarchical relationship, and the hierarchical grouping relationship includes multiple levels, wherein each level corresponds to multiple resource nodes, and the number of groups at different levels is different, and each group at the same level corresponds to the same number of resource nodes. The second processing unit 902 can execute step 302 in the above method embodiment.

Optionally, when the number of resource nodes corresponding to a group at any level is greater than 2, at least two resource nodes corresponding to the group constitute a communication ring.

Optionally, the sum of bandwidths of communication rings corresponding to different groups at the same level is the same, and the bandwidth of the communication ring is the sum of communication bandwidths between resource nodes constituting the communication ring.

Optionally, each group at a different level contains a power-of-two number of resource nodes.

Optionally, the information of each level in the hierarchical grouping relationship is recorded in the form of a linked list, and the linked list of each level includes the node identifiers of the resource nodes contained in each group of the level, and the node identifiers of the resource nodes in different groups are different.

Optionally, the hierarchical grouping relationship is recorded in the form of a binary tree, where each group at the end level of the binary tree includes a node identifier of a resource node, and the node identifiers in different groups are different; the groups at the upper levels include the number of resource nodes corresponding to the groups at the upper levels, wherein the upper levels are any level above the end level.

In the embodiment of the present application, the operations performed by each unit in the resource management device 90 are similar to those described in the embodiments shown in the aforementioned Figures 3 to 8B, and will not be repeated here.

As shown in FIG. 10 , a structure of a resource management device 100 provided in an embodiment of the present application includes:

The receiving unit 1001 is configured to receive a resource request, wherein the resource request includes the number of requested resource nodes. The receiving unit 1001 may execute step 601 in the above method embodiment.

The first processing unit 1002 is configured to obtain at least two sub-requests according to the resource request received by the receiving unit 1001, wherein the sum of the number of resource nodes corresponding to the at least two sub-requests is the number of resource nodes corresponding to the resource request. The first processing unit 1002 may execute step 602 in the above method embodiment.

The second processing unit 1003 is used to allocate a corresponding number of resource nodes to each sub-request obtained by the first processing unit 602 according to the resource management information, and the resource management information includes a hierarchical grouping relationship of multiple resource nodes; wherein the number of resource nodes corresponding to different sub-requests matches the number of resource nodes corresponding to one or more groups at the same level or different levels in the hierarchical grouping relationship.

Optionally, the hierarchical grouping relationship includes multiple levels, wherein each level corresponds to multiple resource nodes, the number of groups in different levels is different, and each group in the same level corresponds to the same number of resource nodes.

Optionally, each grouping at a different level corresponds to an exponential power of 2 resource nodes.

Optionally, each sub-request corresponds to a power-of-two resource node.

Optionally, at least two sub-requests include a first sub-request, and a second processing unit 1003 is specifically used to allocate a resource node corresponding to a node identifier of a group to the first sub-request from a linked list of the target level according to the number of resource nodes corresponding to the first sub-request; wherein the number of node identifiers in each group of the target level is the same as the number of resource nodes corresponding to the first sub-request.

Optionally, at least two sub-requests include a first sub-request, and the second processing unit 1003 is specifically used to allocate resource nodes corresponding to node identifiers of multiple groups to the first sub-request from a linked list of one or more levels according to the number of resource nodes corresponding to the first sub-request; wherein the sum of the numbers of resource nodes corresponding to the node identifiers in the multiple groups is the same as the number of resource nodes corresponding to the first sub-request, the multiple groups belong to the same level, or at least two of the multiple groups belong to different levels.

Optionally, at least two sub-requests include a second sub-request, and the second processing unit 1003 is specifically used to allocate resource nodes corresponding to the node identifier of the last level associated with the target group to the second sub-request according to the number of resource nodes corresponding to the second sub-request, and the target group is a group of the target level of the binary tree; wherein the number of node identifiers of the last level associated with each group of the target level is the same as the number of resource nodes corresponding to the second sub-request.

Optionally, at least two sub-requests include a second sub-request, and the second processing unit 1003 is specifically used to allocate resource nodes corresponding to the node identifiers of the end-level associated with multiple target groups to the second sub-request according to the number of resource nodes corresponding to the second sub-request, and the multiple target groups belong to the same level, or at least two of the multiple groups belong to different levels; wherein the node identifiers of the end-level associated with the multiple target groups are not repeated, and the sum of the number of node identifiers of the end-level associated with the multiple target groups is the same as the number of resource nodes corresponding to the second sub-request.

Optionally, when at least two sub-requests are each of the resource nodes corresponding to a number that satisfies an exponential power of 2, the resource request obtains the minimum number of sub-requests.

The hierarchical grouping relationship is obtained by dividing a plurality of resource nodes according to the levels. When the number of resource nodes corresponding to a group at any level is greater than 2, at least two resource nodes corresponding to the group constitute a communication ring.

Optionally, the sum of the bandwidths of the communication rings corresponding to different groups at the same level is the same, and the bandwidth of the communication ring is the sum of the communication bandwidths between the resource nodes constituting the communication ring.

Optionally, the sum of the bandwidths of communication rings at the same level is the largest of at least two first bandwidth sums, wherein the first bandwidth sum is the sum of the bandwidths of the communication rings obtained when multiple resource nodes are grouped in different ways according to the number of resource nodes corresponding to the same level.

Optionally, the resource node is located in a cloud service system, and the resource node is a computing device card or a virtual machine in the cloud service system.

Optionally, the resource node is located in a data center, the resource node is a computing node of the data center, and the computing node is a server or a virtual machine.

Optionally, the resource node is located in a network system, and the resource node is a network device of the network system, and the network device includes at least one of a firewall, a router, and a switch.

In the embodiment of the present application, the operations performed by each unit in the resource management device 100 are similar to those described in the embodiments shown in the aforementioned Figures 3 to 8B, and will not be repeated here.

In another embodiment of the present application, a computer-readable storage medium is also provided, in which computer execution instructions are stored. When the processor of the resource management device executes the computer execution instructions, the resource management device executes the steps performed by the resource management device in Figures 3 to 8B above.

In another embodiment of the present application, a computer program product is provided. The computer program product includes a computer program code. When the computer program code is executed on a computer, the computer device executes the steps executed by the resource management apparatus in FIG. 3 to FIG. 8B .

In another embodiment of the present application, a chip system is also provided, which includes one or more interface circuits and one or more processors; the interface circuit and the processor are interconnected through lines; the interface circuit is used to receive signals from the memory of the resource management device and send signals to the processor, and the signals include computer instructions stored in the memory; when the processor executes the computer instructions, the terminal executes the steps performed by the resource management device in the above-mentioned Figures 3 to 8B. In one possible design, the chip system may also include a memory, which is used to store program instructions and data necessary for the control device. The chip system may be composed of chips, or may include chips and other discrete devices.

In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above integrated units may be implemented in whole or in part by software, hardware, firmware, or any combination thereof.

When the integrated unit is implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid state drive (SSD)), etc.

Claims

A resource management method, characterized by comprising:

receiving a resource request, the resource request including a number of requested resource nodes;

Obtain at least two sub-requests according to the resource request, the sum of the numbers of resource nodes corresponding to the at least two sub-requests being the number of resource nodes corresponding to the resource request;

According to resource management information, a corresponding number of resource nodes are allocated to each sub-request, and the resource management information includes a hierarchical grouping relationship of multiple resource nodes; wherein the number of resource nodes corresponding to different sub-requests matches the number of resource nodes corresponding to one or more groups at the same level or different levels in the hierarchical grouping relationship.
The method according to claim 1 is characterized in that the hierarchical grouping relationship includes multiple levels, wherein each level corresponds to the multiple resource nodes, the number of groups in different levels is different, and the number of resource nodes corresponding to each group in the same level is the same.
The method according to claim 2 is characterized in that each grouping of the different levels corresponds to an exponential power of 2 resource nodes.
The method according to claim 3 is characterized in that each sub-request corresponds to an exponential power of 2 resource nodes.
The method according to any one of claims 1-4 is characterized in that the information of each level in the hierarchical grouping relationship is recorded in the form of a linked list, and the linked list of each level includes the node identifiers of the resource nodes contained in each group of the level, and the node identifiers of the resource nodes in different groups are different.
The method according to claim 5, wherein the at least two sub-requests include a first sub-request, and the allocating a corresponding number of resource nodes to each sub-request according to the resource management information comprises:

According to the number of resource nodes corresponding to the first sub-request, a resource node corresponding to the node identifier of a group is allocated to the first sub-request from the linked list of the target layer; wherein the number of resource nodes corresponding to the node identifier in each group of the target layer is the same as the number of resource nodes corresponding to the first sub-request.
The method according to claim 5, wherein the at least two sub-requests include a first sub-request, and the allocating a corresponding number of resource nodes to each sub-request according to the resource management information comprises:

According to the number of resource nodes corresponding to the first sub-request, resource nodes corresponding to the node identifiers of multiple groups are allocated to the first sub-request from a linked list of one or more levels; wherein the sum of the numbers of resource nodes corresponding to the node identifiers in the multiple groups is the same as the number of resource nodes corresponding to the first sub-request, the multiple groups belong to the same level, or at least two of the multiple groups belong to different levels.
The method according to any one of claims 1 to 4, characterized in that the hierarchical grouping relationship is recorded in the form of a binary tree, each group at the end level of the binary tree includes a node identifier of a resource node, and the node identifiers in different groups are different;

The grouping of the upper level includes the number of resource nodes corresponding to the grouping of the upper level, wherein the upper level is any level above the last level.
The method according to claim 8, wherein the at least two sub-requests include a second sub-request, and the allocating a corresponding number of resource nodes to each sub-request according to the resource management information comprises:

According to the number of resource nodes corresponding to the second sub-request, the resource node corresponding to the node identifier of the last level associated with the target group is allocated to the second sub-request, and the target group is a group of the target level of the binary tree; wherein the number of node identifiers of the last level associated with each group of the target level is the same as the number of resource nodes corresponding to the second sub-request.
The method according to claim 8, wherein the at least two sub-requests include a second sub-request, and the allocating a corresponding number of resource nodes to each sub-request according to the resource management information comprises:

According to the number of resource nodes corresponding to the second sub-request, resource nodes corresponding to the node identifiers of the most-end level associated with multiple target groups are allocated to the second sub-request, the multiple target groups belong to the same level, or at least two of the multiple groups belong to different levels; wherein the node identifiers of the most-end level associated with the multiple target groups are not repeated, and the sum of the number of node identifiers of the most-end level associated with the multiple target groups is the same as the number of resource nodes corresponding to the second sub-request.
The method according to any one of claims 1-10 is characterized in that the at least two sub-requests are the minimum number of sub-requests obtained by the resource request when the number of resource nodes corresponding to each sub-request satisfies the exponential power of 2.
The method according to any one of claims 1-11 is characterized in that the hierarchical grouping relationship is obtained by dividing the multiple resource nodes in layers, and when the number of resource nodes corresponding to a group at any level is greater than 2, at least two resource nodes corresponding to the group constitute a communication ring.
The method according to claim 12 is characterized in that the sum of the bandwidths of the communication rings corresponding to different groups at the same level is the same, and the bandwidth of the communication ring is the sum of the communication bandwidths between the resource nodes constituting the communication ring.
The method according to claim 12 or 13 is characterized in that the sum of the bandwidths of the communication rings at the same level is the largest of at least two first bandwidth sums, wherein the first bandwidth sum is the sum of the bandwidths of the communication rings obtained when the multiple resource nodes are combined in different ways according to the number of resource nodes corresponding to the groupings at the same level.
The method according to any one of claims 1-14 is characterized in that the resource node is located in a cloud service system, and the resource node is a computing device card or a virtual machine in the cloud service system.
The method according to any one of claims 1-14 is characterized in that the resource node is located in a data center, the resource node is a computing node of the data center, and the computing node is a server or a virtual machine.
The method according to any one of claims 1-14 is characterized in that the resource node is located in a network system, the resource node is a network device of the network system, and the network device includes at least one of a firewall, a router and a switch.
A resource management method, characterized by comprising:

Acquire a topological relationship between multiple resource nodes, where the topological relationship represents a communication relationship between the multiple resource nodes;

According to the topological relationship between the multiple resource nodes, the multiple resource nodes are grouped according to levels to obtain a hierarchical grouping relationship of the multiple resource nodes. The hierarchical grouping relationship is obtained by dividing the multiple resource nodes according to levels. The hierarchical grouping relationship includes multiple levels, wherein each level corresponds to the multiple resource nodes, different levels have different numbers of groups, and each group at the same level corresponds to the same number of resource nodes.
The method according to claim 18 is characterized in that when the number of resource nodes corresponding to a group at any level is greater than 2, at least two resource nodes corresponding to the group constitute a communication ring.
The method according to claim 19 is characterized in that the sum of the bandwidths of the communication rings corresponding to different groups at the same level is the same, and the bandwidth of the communication ring is the sum of the communication bandwidths between the resource nodes constituting the communication ring.
The method according to claim 19 or 20 is characterized in that the sum of the bandwidths of the communication rings at the same level is the largest of at least two first bandwidth sums, wherein the first bandwidth sum is the sum of the bandwidths of the communication rings obtained when the multiple resource nodes are combined in different ways according to the number of resource nodes corresponding to the groupings at the same level.
The method according to any one of claims 18-21 is characterized in that each group of the different levels contains an exponential power of 2 resource nodes.
The method according to any one of claims 18-22 is characterized in that the information of each level in the hierarchical grouping relationship is recorded in the form of a linked list, and the linked list of each level includes the node identifiers of the resource nodes contained in each group of the level, and the node identifiers of the resource nodes in different groups are different.
The method according to any one of claims 18 to 22, characterized in that the hierarchical grouping relationship is recorded in the form of a binary tree, each group at the end level of the binary tree includes a node identifier of a resource node, and the node identifiers in different groups are different;

The grouping of the upper level includes the number of resource nodes corresponding to the grouping of the upper level, wherein the upper level is any level above the last level.
A resource management device, characterized in that it includes: a communication interface, a processor and a memory, wherein the communication interface and the processor are coupled to the memory, and the memory is used to store programs or instructions. When the program or instructions are executed by the processor, the client executes the method described in any one of claims 1-17 or 18-24.
A computer-readable storage medium, characterized in that instructions are stored in the computer-readable storage medium, and when the instructions are executed on a computer device, the computer device executes the method as described in any one of claims 1-17 or 18-24.
A computer program product, characterized in that the computer program product comprises computer program code, and when the computer program code is run on a computer device, the computer device executes the method as described in any one of claims 1-17 or 18-24.
A chip system, characterized in that the chip system includes one or more interface circuits and one or more processors; the interface circuit and the processor are interconnected through lines; the interface circuit is used to receive signals from a client's memory and send signals to the processor, and the signals include computer instructions stored in the memory; when the processor executes the computer instructions, the chip system executes the method described in any one of claims 1-17 or 18-24.
A resource management system, characterized in that it includes a resource management device and multiple resource nodes, and the resource management device is used to execute the method described in any one of claims 1-17 or 18-24.