KR102088286B1 - Method for managing and sharing symmetric flow and asymmetric flow in duplexed network - Google Patents

Abstract

The network redundancy device according to the present invention stores information on the flow when information of a flow introduced to it is not generated, and provides information on the flow to the master unit when the flow corresponds to an asymmetric flow A plurality of network interface units and a master unit that stores information of a flow provided from the network interface unit and provides information of the opposite flow to the network interface unit when information of the opposite flow for the flow is stored. It is composed. Therefore, when using the present invention, it is possible to perform integrated management and analysis for asymmetric flows in which the inflow path and the outflow path of packets are configured differently.

Description

Method for managing and sharing symmetric flow and asymmetric flow in duplexed network}

The present invention relates to network duplexing, and more specifically, for an asymmetric flow in which a packet inflow path to a network duplexing device and a packet outflow path from a network duplexing device are configured differently, integrated It relates to a method and apparatus for enabling flow information management and analysis.

When network redundancy is applied in a point of presence (PoP) branch of an operator or an enterprise network, two or more traffic inflow paths or outflow paths may be configured. In this case, a situation may arise in which a traffic from a node (terminal) corresponding to a management network and a connection path from a remote location uses different paths.

When the flow of traffic generated from the remote site to the terminal and the flow of traffic generated from the terminal to the remote site take different paths in the network redundancy device, such a flow may be defined as an asymmetric flow. . On the other hand, when the flow of traffic generated from a remote site to the terminal and the flow of traffic generated from the terminal to the remote site take the same path in the network redundancy device, such a flow may be defined as a symmetric flow.

At this time, since the flows constituting the asymmetric flow are managed in different network interface units, unified management of the flows leads to a difficult problem, which is consistent with service provision or quality of service (QoS) in a network redundancy environment. It makes the management difficult.

Therefore, if a unified information management method for such an asymmetric flow is provided in a redundant network having multiple interfaces, it is possible to make advanced responses such as service provision and QoS.

For network service providers, enterprise network administrators, and Telco operators, if the ability to analyze packets flowing into all interfaces is provided under the same conditions, scalability, auto-load balancing, and additional capacity It will be able to provide various functions such as (Additional Capacity) and Flexible Deployment.

An object of the present invention for solving the above problems is to provide a network redundancy device that enables integrated analysis and management for asymmetric flows having different inflow paths and outflow paths of packets.

Another object of the present invention to solve the above problems is to provide a network redundancy method that enables integrated analysis and management for asymmetric flows having different inflow and outflow paths of packets.

The present invention for achieving the above object is a network duplication apparatus having a master unit and a plurality of network interface units, and stores information of the flow when the information of the flow introduced to itself is not generated, and the flow is When it corresponds to an asymmetric flow, the flow of information provided from the plurality of network interface units and the network interface unit is provided to provide the information of the flow to the master unit, and to request and receive information of the flow opposite to the flow to the master unit And a master unit that stores information of, checks whether information of the opposite flow for the flow is stored, and provides information of the opposite flow to the network interface unit when the information of the opposite flow is stored, The plurality of network interface units is composed of network interface cards mounted in a slot of one board, and the master unit provides a network redundancy device including a central processing unit (CPU).

Here, the network interface unit and the master unit may be configured to interface with a PCI-Express interface method.

Here, the network redundancy device may operate based on a connection-oriented protocol. In this case, if the SYN-ACK packet flows in, the network interface unit corresponds to the flow corresponding to the asymmetric flow if information on the flow to which the SYN packet corresponding to the SYN-ACK packet belongs is not stored in the network interface unit. Can be recognized. At this time, when the ACK packet for the SYN-ACK packet flows in, the network interface unit recognizes that the flow corresponds to an asymmetric flow if the information of the flow to which the SYN-ACK packet belongs is not stored in the network interface unit. can do.

Here, the network redundancy device may operate based on a connectionless protocol. At this time, the network interface unit may recognize that the flow corresponds to an asymmetric flow if the flow opposite to the flow does not flow into the network interface unit for a predetermined period.

Here, the network interface unit provides traffic information of flows corresponding to the asymmetric flow to the master unit at the time of processing traffic of the flows corresponding to the asymmetric flow or at a predetermined period, and the master unit is the network interface unit It may be configured to store traffic information of flows corresponding to the asymmetric flow provided from.

Here, the network interface unit is a control interface unit for interfacing with the master unit, a network interface unit for transmitting and receiving the flow to and from the outside, a local flow header table storage unit for storing information on flows processed by the network interface unit When the information of the flow corresponding to the packet is not generated by analyzing the traffic flowing through the traffic information storage unit storing the traffic information of the flows and the network interface unit, the flow of the flow is generated, and the local flow header is generated. Stored in a table storage unit, when the flow corresponds to an asymmetric flow, the flow information is provided to the master unit through the control interface, and the information of the flow opposite to the flow is transferred to the master unit It may be configured to include a control unit for requesting and receiving, and storing the traffic information of the flow in the traffic information storage unit.

Here, the master unit, the control interface unit for interfacing with the network interface unit, an asymmetric flow header table storage unit for storing information of an asymmetric flow received from the network interface unit, traffic for storing traffic information of the asymmetric flows When the information of the opposite flow is stored by referring to the information of the asymmetric flow received from the network interface unit through the information storage unit and the control interface unit, the information of the opposite flow is stored. It may be configured to provide a control unit for providing information of the opposite flow to the network interface unit, and storing the information of the flow.

The present invention for achieving the above other object, the master network unit including a CPU (Central Processing Unit) and a network duplication apparatus having a plurality of network interface units consisting of network interface cards mounted in a slot of one board As an operation method, the network interface unit analyzes the flow flowing into the network interface unit, the network interface unit storing information of the flow when the information of the flow is not generated, the network The interface unit determines whether the flow corresponds to an asymmetric flow, and when the flow is an asymmetric flow, provides information of the flow to the master unit, and requests information of the flow opposite to the flow to the master unit Receiving, the master unit storing information of the flow provided from the network interface unit, and the master unit confirming whether information of the opposite flow for the flow is stored, and information of the opposite flow is stored Provides a network redundancy method including providing information of the opposite flow to the network interface unit, if present.

Here, the network interface unit and the master unit may be configured to interface with a PCI-Express interface method.

Here, the flow may be a flow according to a connection-oriented protocol. In this case, if the SYN-ACK packet flows in, the network interface unit corresponds to the flow corresponding to the asymmetric flow if information on the flow to which the SYN packet corresponding to the SYN-ACK packet belongs is not stored in the network interface unit. Can be recognized. At this time, when the ACK packet for the SYN-ACK packet flows in, the network interface unit recognizes that the flow corresponds to an asymmetric flow if the information of the flow to which the SYN-ACK packet belongs is not stored in the network interface unit. can do.

Here, the flow may be a flow according to a non-connection oriented protocol. At this time, the network interface unit may recognize that the flow corresponds to an asymmetric flow if the flow opposite to the flow does not flow into the network interface unit for a predetermined period.

Here, the network redundancy method further includes the step of providing, by the network interface unit, traffic information of flows corresponding to the asymmetric flow to the master unit at a time of traffic processing of flows corresponding to the asymmetric flow or at a predetermined cycle. It can contain. In this case, the network redundancy method may further include the step of storing traffic information of flows corresponding to the asymmetric flow provided by the master unit from the network interface unit.

In general, network equipment performs flow-based analysis distributed through reflection of a flow table only for traffic flowing into a corresponding interface. In this case, flow-based analysis cannot be accurately provided in a redundant or multi-path network environment.

When applying the clustering technique for asymmetric flows using the PCIe NIC card proposed in the present invention, asymmetric flows are integratedly managed for multiple paths related to in-out and traffic information of asymmetric flows is also managed in real time. I can do it.

Therefore, when the present invention is used, traffic analysis such as flow-based deep packet inspection is possible, and flow-based application recognition and control are possible.

1 is a conceptual diagram illustrating the concept of a network redundancy device according to the present invention.
2 is a block diagram illustrating an exemplary configuration of a network redundancy apparatus according to the present invention.
3 is a block diagram for explaining an example of a configuration of a network interface unit of a network redundancy device according to the present invention.
4 is a block diagram for explaining a configuration example of a master unit of a network redundancy device according to the present invention.
5 is a conceptual diagram illustrating packet transmission between a client and a server according to the TCP protocol in order to explain the operation of the network duplication apparatus according to the present invention for a connection-oriented protocol.
6 is a message flow chart for illustrating an operation procedure for a connection-oriented protocol between components of a network redundancy device according to the present invention.
7 is a conceptual diagram illustrating packet transmission between a client and a server according to a non-TCP protocol in order to describe an operation for a non-connection-oriented protocol of a network duplication apparatus according to the present invention.
8 is a message flow diagram for illustrating an operating procedure for a non-connection oriented protocol between components of a network redundancy device according to the present invention.

The present invention can be applied to various changes and may have various embodiments, and specific embodiments will be illustrated in the drawings and described in detail in the detailed description. However, this is not intended to limit the present invention to specific embodiments, and should be understood to include all modifications, equivalents, and substitutes included in the spirit and scope of the present invention. In describing each drawing, similar reference numerals are used for similar components.

Terms such as first, second, A, and B can be used to describe various components, but the components should not be limited by the terms. The terms are used only for the purpose of distinguishing one component from other components. For example, the first component may be referred to as a second component without departing from the scope of the present invention, and similarly, the second component may be referred to as a first component. The term and / or includes a combination of a plurality of related described items or any one of a plurality of related described items.

When an element is said to be "connected" or "connected" to another component, it is understood that other components may be directly connected to or connected to the other component, but other components may exist in the middle. It should be. On the other hand, when a component is said to be "directly connected" or "directly connected" to another component, it should be understood that no other component exists in the middle.

The terms used in this application are only used to describe specific embodiments, and are not intended to limit the present invention. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this application, the terms "include" or "have" are intended to indicate the presence of features, numbers, steps, actions, components, parts or combinations thereof described herein, one or more other features. It should be understood that the existence or addition possibilities of fields or numbers, steps, operations, components, parts or combinations thereof are not excluded in advance.

Unless defined otherwise, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by a person skilled in the art to which the present invention pertains. Terms such as those defined in a commonly used dictionary should be interpreted as having meanings consistent with meanings in the context of related technologies, and should not be interpreted as ideal or excessively formal meanings unless explicitly defined in the present application. Does not.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a conceptual diagram illustrating the concept of a network redundancy device according to the present invention.

Referring to FIG. 1, the network redundancy device 110 provides services for packets flowing in and out through four nodes 101, 102, 103, and 104. At this time, the nodes 101 and 102 may be understood to be connected to the IP backbone network 120, and the nodes 103 and 104 may be understood to be connected to the subscriber access network 130. In addition, in the following description, it is assumed that the server 121 is located in the IP backbone network and the client 131 is located in the subscriber access network.

The interfaces 111, 112, 113, and 114 between the nodes 101, 102, 103, and 104 may include multiple strands of 1 gigaline or 10 gigabit lines, and the interface between each node is a network interface. Corresponds to a network interface controller (NIC). That is, the network interface unit of the network redundancy device according to the present invention, which will be described later, corresponds to a network interface controller that is responsible for the interface between each node.

The traffic flowing from the server 121 to the node 101 through the Internet line connected through the link 141 is a client 131 using the node 103 or the node 104 where the corresponding client 131 is located. Is delivered to.

If the traffic of the client 131 flows into the node 103 through the connected line of the link 143, the traffic from the client is a node by a function such as load balancing according to the network state of the network and the processing equipment having a session recognition function. It is delivered to the server 121 through the 101 or the node 102.

That is, if the traffic delivered from the server 121 to the client 131 flows into the node 101 and passes through the node 103, the traffic transmitted from the client to the server flows into the node 103 and enters the node 101 ) Or may be delivered to the server 121 through the node 102.

If the traffic from the client to the server flows from the client to the node 103 and passes to the server through the node 101 (ie, the traffic from the client to the server and the traffic from the server to the client take the same path) If), the flow from the client to the server and the flow from the server to the client can be defined as a symmetric flow.

On the other hand, if the traffic from the client to the server flows from the client to the node 103 and is delivered to the server through the node 102, the flow from the client to the server and the flow from the server to the client are asymmetric flows. Can be defined.

The network redundancy apparatus according to the present invention is characterized in that it is possible to integrally manage information about asymmetric flows and traffic information of asymmetric flows in such a redundant network structure.

This enables network service providers, enterprise network administrators, and Telco operators to manage flow information for asymmetric flows and traffic information for flows in an integrated manner, so that packets flowing into all interfaces can be analyzed under the same conditions. Can provide functionality.

Asymmetric Flow  Configuration of network redundancy device for management

2 is a block diagram illustrating an exemplary configuration of a network redundancy apparatus according to the present invention.

Referring to FIG. 2, the network redundancy device 200 according to the present invention includes a plurality of network interface units 210, 220, 230, and 240 and at least one master unit 250 connected to the plurality of network interface units. It may be configured to include. For example, the network interface units 210, 220, 230, and 240 may each be configured in the form of a PCI-Express type network interface card (NIC) mounted in a slot on one board. The master unit may include a main CPU 255 (eg, x86 CPU).

In one embodiment, the network redundancy device according to the present invention has four network interface units 210, 220, 230, each having a network interface (eg, SFP + interface; 10 Gigabit Ethernet; 211, 221, 231, 241). 240) and a master unit 250 including a CPU 255 connected to network interface units. Since the master unit may not be implemented in a separate card form, and the main function of the master unit 250 may be implemented using the main CPU 255, the cost of implementation may be reduced. Therefore, the network interface units and the master unit may be configured on one server board.

The network interface units and the master unit may be configured to perform mutual communication through certain high-speed interfaces 261, 262, 263, and 264. At this time, the network interface units and the master unit may communicate with each other through an interface bus of slots in which the units are mounted.

In the network redundancy device according to the present invention, network interface units are responsible for transmitting and receiving traffic packets to and from the outside, analyzing and transmitting packets transmitted and received through themselves to store and manage information of flows passing through them and traffic information of the flows. However, each network interface unit cannot have information on the opposite flow constituting the asymmetric flow when the flow passing through it corresponds to the asymmetric flow. Therefore, information on the asymmetric flow is managed integrally by the master unit.

3 is a block diagram for explaining an example of a configuration of a network interface unit of a network redundancy device according to the present invention.

Referring to FIG. 3, one configuration example 300 of the network interface unit of the network duplication device according to the present invention includes a control interface unit 310, a control unit 320, and a local flow header table (LFHT) storage. It may be configured to include a unit 330, a flow traffic information storage unit 340 and the network interface unit 350.

First, the control interface 310 is a component for interfacing with the master unit described above. At this time, the control interface 310 may communicate with the master unit through an interface bus of slots in which the network interface units are mounted. The control interface unit may be connected to the master unit in a PCI-Express interface method as an embodiment. The master unit may also include a control interface unit 410 corresponding to the control interface unit 310, and the configuration of the master unit will be described later.

The network interface unit 350 is a component for processing a high-speed network interface such as, for example, an SFP + interface (10 Gigabit Ethernet), and is a component for performing a function of a conventional NIC.

Meanwhile, the control unit 320, the local flow header table storage unit 330, and the flow traffic information storage unit 340 correspond to core components of the network interface unit for performing the operation of the network duplication apparatus according to the present invention.

The local flow header table storage unit 330 is a component that stores a local flow header table (LFHT), and the local flow header table (LFHT) is a component that records information about flows passing through each network interface unit. to be. Flows in which the information is recorded in the local flow header table may be both flows constituting a symmetric flow and one flow constituting an asymmetric flow.

The flow traffic information storage unit 340 is a component that stores traffic information of flows stored in the local flow header table storage unit. Here, the traffic information of the flow may include, for example, the number of packets of the flow, the total amount of data, the delivery speed, and the number of hops. Depending on the embodiment, the flow traffic information storage unit 330 and the local flow header table storage unit 340 may be integrally configured.

The control unit 320 analyzes the traffic packets flowing into the network interface unit 350, and when the information of the first flow corresponding to the traffic packet is not generated in the LFHT, generates the information of the first flow to generate the LFHT It serves to save on.

In addition, when the controller 320 recognizes that the flow opposite to the first flow (hereinafter, the second flow) is processed by a network interface unit other than itself (that is, when the first flow is recognized as belonging to an asymmetric flow) , The information of the first flow is provided to the master unit 250 through the control interface 310. In addition, the control unit 320 may request and receive information of the second flow, which is the flow opposite to the first flow, to the master unit.

At this time, the method in which the control unit recognizes that the flow opposite to the first flow (hereinafter, the second flow) is processed in another network interface unit is in the case of a flow according to a connection-oriented protocol and a flow according to a non-connection-oriented protocol. It can be configured differently.

The operations of the above-described control unit 320 and the roles of the local flow header table storage unit 330 and the flow traffic information storage unit 340 will be described later with reference to specific operation examples.

4 is a block diagram for explaining a configuration example of a master unit of a network redundancy device according to the present invention.

Referring to FIG. 4, one configuration example 400 of the master unit of the network duplication apparatus according to the present invention includes a control interface unit 410, a CPU 420, and an asymmetric flow header table (AFHT) storage unit 430 and an asymmetric flow traffic information storage unit 440.

First, the control interface unit 410 is a component for interfacing with the above-described network interface units. At this time, the control interface unit 410 may communicate with the network interface units through an interface bus of slots in which the network interface units are mounted. The control interface unit may be connected to the network interface units described above in a PCI-Express interface method as an embodiment.

Meanwhile, the CPU 420, the asymmetric flow header table storage unit 430, and the asymmetric flow traffic information storage unit 440 correspond to core components of the master unit for performing the operation of the network duplication apparatus according to the present invention.

The asymmetric flow header table storage unit 430 is a component that stores the asymmetric flow header table AFHT, and the asymmetric flow header table AFHT stores information of asymmetric flows. That is, in the local flow header table of network interface units, symmetric flows belonging to the network interface units and one side flows of the asymmetric flows passing through the network interface units are recorded, while the network interface units are in the asymmetric flow header table. Information on all asymmetric flows to be processed is recorded integrally.

The asymmetric flow traffic information storage unit 440 is a component that stores traffic information of asymmetric flows stored in the asymmetric flow header table storage unit. Here, the traffic information of the asymmetric flow may include, for example, the number of packets of the asymmetric flow, the total amount of data, the transmission rate, the number of hops, and the like. Traffic information of the asymmetric flow is collected from network interface units that process one side of the flow constituting the asymmetric flow.

Depending on the embodiment, the asymmetric flow traffic information storage unit 440 and the asymmetric flow header table storage unit 430 may be configured integrally. The asymmetric flow traffic information storage unit 440 and the asymmetric flow header table storage unit 430 may include volatile memory (eg, RAM, etc.), non-volatile memory (eg, FLASH memory, etc.) or hard disk. It can be located on various storage devices such as drives.

The CPU 420 receives information about the asymmetric flow from the network interface unit and stores it in the asymmetric flow header table, and information on the other side of the flow of one side of the flow belonging to the asymmetric flow received from the network interface unit is already stored in the asymmetric flow header table. If stored, it acts as a control to return it to the network interface unit. That is, the CPU 420 executes program code for performing the above-described operation.

The above-described operation of the CPU 420 and the roles of the asymmetric flow header table storage unit 430 and the asymmetric flow traffic information storage unit 440 will be described later with reference to specific operation examples.

Asymmetric Flow  Network redundancy method for management

Hereinafter, the operation of the network redundancy device according to the present invention will be described for each of the connection-oriented protocol and the non-connection-oriented protocol. The connection-oriented protocol may include, for example, TCP (Transfer Control Protocol), and the non-connection-oriented protocol may include non-TCP protocol, such as User Datagram Protocol (UDP) or Internet Control Message Protocol (ICMP). .

1) Operation method corresponding to connection-oriented protocol

FIG. 5 is a conceptual diagram illustrating packet transmission between a client and a server according to the TCP protocol in order to explain the operation of the network duplication apparatus according to the present invention, and FIG. 6 is a component of the network duplication apparatus according to the present invention. It is a message flow chart to illustrate the operation procedure for the connection-oriented protocol between them.

First, referring to FIG. 5, the client 510 sends a Synchronization (SYN) packet CP1 to the server 520 to request the creation of a TCP flow, and the server 520 responds to the TCP flow generation This is illustrated. Of course, it is also possible to request the creation of a TCP flow by sending a SYN packet from the server side to the client side.

In the TCP protocol, the SYN-ACK packet is transmitted by the other side (e.g., the server) to the SYN packet sent by one side (e.g., the client), and the other side of the SYN packet is sent back to the SYN-ACK packet. By sending, two flows can be created: the flow from the client to the server and the flow from the server to the client.

In the following description, packets transmitted from the client side to the server side are indicated by CP # n, and packets transmitted from the server side to the client side are indicated by SP # n. Here, #n corresponds to an expression for specifying the order of packets.

At this time, 'client' and 'server' are examples used for convenience only to distinguish the subject of packet transmission and reception, and do not define the roles of the client and the server in a strict sense. It should be noted that it is not limited to. For example, in the following description, the client and the server may be referred to as a first device (terminal) and a second device (terminal), respectively.

Hereinafter, an operation procedure of the network duplication apparatus according to the present invention will be described with reference to FIGS. 5 and 6 in parallel.

In the following example, the flow from the client to the server (hereinafter referred to as the client flow or "C->S" flow) is processed through the first network interface unit, but the flow from the server to the client (hereinafter, the server) Flow or “S-> C” flow) is described on the assumption that the situation is processed through the third network interface unit, that is, an asymmetric flow has occurred.

When CP1 (Client Packet 1) first arrives at the first network interface unit 210 (601), the first network interface unit 210 searches for its internal symmetric flow header table (LFHT) and corresponds to CP1. After confirming that the flow has not been created, a client flow 602 is newly added to the LFHT. That is, at this time, information indicating that the flow from the client to the server is processed through the first network interface unit is stored in the LFHT of the first network interface unit. Exemplarily, the information of the flow is stored in the LFHT in a notation manner such as "C-> S" (in FIG. 6, the shadow of the LFHT is for indicating an item currently processed on the LFHT).

At this time, since CP1 is a SYN packet, the first network interface unit still transmits a SYN-ACK packet (ie, traffic from the server to the client; SP1, which will be described later) corresponding to the SYN packet through the first network interface unit. It is not yet possible to recognize whether it will be transmitted through a network interface unit other than itself. Therefore, the first network interface unit generates only the flow corresponding to CP1 in its LFHT and does not make a separate request to the master unit 210.

Next, when the server packet (SP1) first arrives at the third network interface unit 230 (603), the third network interface unit 230 searches for its internal LFHT, and the flow corresponding to the SP1 is still generated. After confirming that it has not been done, a server flow (604) is newly added to the LFHT. That is, at this time, information indicating that the flow from the server to the client is processed through the third network interface unit is stored in the LFHT of the third network interface unit. For example, it is expressed that the information of the flow is stored in the LFHT in a notation manner such as "S-> C".

At this time, the control unit of the third interface unit requests the master unit to register the "S-> C" flow as an asymmetric flow through the control interface unit. The third network interface unit recognized that SP1 is a SYN-ACK packet (check the indicator of the packet header), but it knows that the SYN packet corresponding to the SYN-ACK packet has never been processed by the third network interface unit (LFHT phase). Because there is no information of "C-> S" flow), it can be recognized that the SP1 packet corresponds to an asymmetric flow.

The master unit 210 adds the server flow (“S-> C”) information received from the third network interface unit 230 to the AFHT (606), and sends the result to the control unit of the third network interface unit 230. Respond. At this time, the master unit stores information indicating that the corresponding flow ("S-> C") is being processed by the third network interface unit in the AFHT. For example, in the AFHT of the master unit, it is recorded that the flow from the server to the client ("S-> C") is processed in the third network interface unit (NIU3) (in FIG. 6, the shadow of AFHT is currently processed on AFHT) To indicate the items to be).

At this time, the master unit may notify the third network interface unit that the other flow is already stored in the AFHT of the master unit, if a record for the opposite flow (“C-> S”) already exists in the AFHT. In the example of FIGS. 5 and 6, since the opposite flow ("C->S") has not yet been created in the AFHT of the master unit, the master unit notifies that the opposite flow is not stored in the master unit. . On the other hand, the master unit implicitly reverses the flow by not responding to a query as to whether or not a record for the flow of the opposite side of the third network interface unit ("C->S") is present in the AFHT ("C-> S ") may be configured to inform the third network interface unit that the AFHT does not exist.

Next, when the CP2 arrives (607), the control unit of the first network interface unit 210 searches for its own LFHT to check whether there is no other flow (“S-> C”), and the client-> server flow (client Flow) is an asymmetric flow, transmits its client flow information to the master unit 210 through the control interface, and requests the opposite flow ("S-> C") information (608). At this time, since the first network interface unit knows that the previous packet CP1 was a SYN packet, the “C-> S” flow is generated only by the fact that the opposite flow (“S-> C”) is not generated in its LFHT. It can be recognized that it is an asymmetric flow.

The master unit receiving the client flow (“C-> S”) information of the first network interface unit and the search request of the opposite flow from the first network interface unit adds the requested client flow information to the AFHT (609), and the opposite side If there is SF, the result is returned to the corresponding AFA 221. At this time, the master unit records in the AFHT that the first network interface unit processes the client flow ("C->S") (609).

Through the above-described procedures, the first network interface unit recognizes that the client flow ("C-> S") it is processing corresponds to an asymmetric flow, and the third network interface unit also processes the server flow (" S-> C ") corresponds to the asymmetric flow.

Therefore, in subsequent procedures, the first network interface unit continuously updates and stores traffic information (eg, the number of packets, the total amount of data, the transmission rate, the number of hops, etc.) of its flow (“C-> S”). The third network interface unit continuously updates and stores traffic information of its flow ("S-> C").

In addition, the first and third network interface units may transmit updated traffic information to the master unit when packets belonging to each flow are transmitted and received. Alternatively, the first and third network interface units may transmit traffic information of each flow to the master unit at a predetermined period. The master unit stores the traffic information of the asymmetric flow received from each network interface unit in the asymmetric flow traffic information storage unit 440.

At this time, the predetermined period may be a predetermined time interval. Alternatively, the first and third network interface units may be configured to transmit traffic information of each flow to the master unit when a predetermined event condition is satisfied.

At this time, the first and third network interface units may be configured to maintain traffic information for flows that are not asymmetric flows only and do not provide them to the master unit. According to the above-described method, the first and third network interface units can distinguish whether the flow passing through them is a symmetrical flow or a flow belonging to an asymmetrical flow, so only the traffic information for the symmetrical flow is stored by itself and provided to the master unit You may not.

This can have the effect of reducing the communication bandwidth burden between the master unit and the network interface units. Meanwhile, the master unit comprehensively manages information on asymmetric flows of the network redundancy device to which it belongs. The master unit may provide asymmetric flow information managed by an external (user / administrator) request to enable integrated management.

Meanwhile, hereinafter, a procedure for processing flow termination in the network redundancy device according to the present invention will be described.

In the third network interface unit, when SPn (Fin packet) arrives (610), the server flow ("S-> C") is deleted (611) from its own LFHT, and the master unit is also requested to end the server flow. (612). Upon receiving the request to end the server flow from the third network interface unit, the master unit deletes the server flow from the AFHT (613) and returns the result to the third network interface unit.

When the CPn (Fin packet) arrives (614) in the first network interface unit, the client flow ("C-> S") is deleted from the LFHT (615), and the master unit is also requested to terminate the client flow. (616). Upon receiving the request to end the client flow from the first network interface unit, the master unit deletes the client flow from the AFHT (617) and returns the result to the first network interface unit.

As a result, in the AFHT of the master unit at the request of the first network interface unit and the third network interface unit, all information about the asymmetric flow is deleted.

2) Operation method corresponding to non-connection oriented protocol

7 is a conceptual diagram illustrating packet transmission between a client and a server according to a non-TCP protocol in order to explain the operation of the network duplication apparatus according to the present invention for a non-connection oriented protocol, and FIG. 8 is a network redundancy according to the present invention It is a message flow diagram to illustrate the operating procedure for a connectionless oriented protocol between components of a device.

First, referring to FIG. 7, a situation in which the client 710 and the server 720 exchange data with each other according to a non-connection-oriented protocol is illustrated.

In the case of the connection-oriented protocol described above (eg, the TCP protocol), there is an explicit flow generation procedure by exchanging SYN, SYN-ACK, and ACK packets between one side and the other side, but in a non-connection-oriented protocol, The flow is implicitly generated by sending and receiving data packets without a process, and if there is no exchange of data packets for a predetermined period, the flow is implicitly released.

In the following description, as in the previous connection-oriented embodiment, packets transmitted from the client side to the server side are indicated by CP # n, and packets transmitted from the server side to the client side are indicated by SP # n. Here, #n corresponds to an expression for specifying the order of packets.

Hereinafter, an operation procedure of the network redundancy device according to the present invention will be described with reference to FIGS. 7 and 8 in parallel.

In the following example, the flow from the client to the server (hereinafter referred to as the client flow or "C->S" flow) is processed through the first network interface unit, but the flow from the server to the client (hereinafter, the server) Flow or “S-> C” flow) is described on the assumption that the situation is processed through the third network interface unit, that is, an asymmetric flow has occurred.

When a CP1 (Client Packet) first arrives at the first network interface unit (801), the first network interface unit searches for its LFHT to confirm that flow information corresponding to CP1 has not been generated, and the client flow to the LFHT ( "C->S") is newly added (802). In this case, the first network interface unit does not know whether the flow (ie, the server flow) on the other side of the flow (client flow) corresponding to CP1 flows into the first network interface unit, so only the client flow is added to its LFHT.

When a server packet (SP1) first arrives at the third network interface unit (803), the third interface unit also searches its own LFHT to confirm that flow information corresponding to SP1 has not been generated, and the server flow to the LFHT (" S-> C ") is newly added (804). In this case, the third network interface unit adds only the server flow to its own LFHT because it cannot yet confirm that the flow opposite the SP1 (server flow) flow (ie, the client flow) has passed through the first network interface unit. do.

Unlike the previous connection-oriented protocol, in the case of a non-connection-oriented protocol, there is no packet indicator defining a packet exchange procedure such as a SYN packet or a SYN-ACK packet, so the third network interface unit also has SP1. The fact that he came to himself cannot judge whether the server flow is an asymmetric flow or not.

Therefore, in the first network interface unit, a predetermined timeout (eg, basic 1 second) time or a period during which a packet of a predetermined minimum number of packets (eg, basic 3) is transmitted / received through the first network interface unit after the CP1 packet If a traffic packet belonging to the opposite flow (i.e., the server flow) does not flow to the first network interface unit (805), the master unit is requested to register the client flow with the AFHT as an asymmetric flow while the other flow (server flow) Request information about (806).

The master unit adds the client flow information received from the first network interface unit to the AFHT (807), and responds the result to the control unit of the first network interface unit. At this time, if the information of the opposite flow (server flow) already exists in the AFHT, the master unit returns the information of the opposite flow to the first network interface unit.

If the other side flow already exists, when the master unit notifies the first network interface unit that the other side flow exists, the first network interface unit clearly recognizes that the client flow is a flow corresponding to the asymmetric flow. However, in the example of FIG. 8, since the server flow has not yet been recorded in AFHT, the master unit explicitly notifies the first network interface unit that the opposite flow (server flow) has not yet been recorded in AFHT, or a separate response. By not doing so, the first network interface unit is implicitly informed that the opposite flow (server flow) has not yet been recorded in the AFHT.

In addition, in the third network interface unit, after a SP1 packet, a predetermined minimum number of packets (e.g., basic 3) is transmitted / received through the third interface unit or during a predetermined timeout (e.g., basic 1 second) time. If a packet belonging to the flow opposite the flow (that is, the client flow) does not flow into the third network interface unit (808), the master unit requests registration of the server flow ("S-> C") to AFHT (809), Request information about the opposite flow ("C-> S") (809).

The master unit adds the server flow information received from the third network interface unit to the AFHT (810) and responds the result to the control unit of the third network interface unit. At this time, if the information of the opposite flow (client flow) already exists in the AFHT, the master unit returns the information of the opposite flow to the third network interface unit.

If the other side of the flow already exists, if the master unit notifies the third network interface unit that the other side of the flow exists, the third network interface unit clearly recognizes that the server flow is a flow corresponding to the asymmetric flow. In the example of Fig. 8, at this point, the client flow ("C->S") is recorded in AFHT, so the master unit explicitly notifies the third network interface unit that the opposite flow (client flow) is recorded in AFHT. can do. Accordingly, the third network interface unit can explicitly recognize that its server flow is a flow corresponding to an asymmetric flow.

The control unit of the first network interface unit continues to process a certain number of packets (for example, basic = 10) while the maximum number of packets (for example, basic = 100) is maintained when the opposite flow (server flow) does not flow into itself. Whenever possible, the updated traffic information of the flow is transmitted to the control unit of the master unit (for example, 811), and a request message for the opposite flow information is transmitted. The master unit adds client flow traffic information received from the first network interface unit to the asymmetric flow traffic information storage unit and responds the result to the control unit of the first network interface unit. At this time, if there is an opposite flow, information on the opposite flow is also returned.

Likewise, the control unit of the third interface unit continues to have a certain number of packets (eg, basic = 10) while the maximum number of packets (eg, basic = 100) is reached if the other side of the flow (client flow) does not continue to flow thereafter. Whenever it is processed, it sends the updated traffic information of the flow to the master unit (eg, 812) and sends a request message for the opposite flow. The master unit adds server flow traffic information received from the third network interface unit to the asymmetric flow traffic information storage unit and responds the result to the control unit of the third network interface unit. At this time, if there is an opposite flow, information on the opposite flow is also returned.

That is, the first interface unit and the third interface unit, if the flows of the flows opposite to the previously identified flows do not appear to the user, the flows are regarded as asymmetric flows, and the traffic information of the flows is transmitted to the master unit at a predetermined cycle. In addition, the first interface unit and the third interface unit can receive information on the other side of the flow from the master unit at any time, and when the other side of the flow exists in another interface unit other than itself, their flows are explicitly asymmetric. It can be recognized that it belongs to the flow.

On the other hand, in the case of a non-connection-oriented protocol, the end of the flow is implicitly determined by the fact that, unlike the connection-oriented protocol, each network interface unit does not flow packets belonging to a flow that it is processing for a predetermined time. . For example, when a packet belonging to its own flow (client flow) does not flow for a predetermined time, the first network interface unit determines that the client flow is ended. Likewise, when a packet belonging to the third network interface unit does not flow for a predetermined period of time, it is determined that the server flow is ended. Hereinafter, a procedure in the case where the first network interface unit and the third network interface unit respectively determine that their flow has ended is described.

First, when the first network interface unit recognizes that the client flow has ended (813), it deletes the client flow information from its LFHT (814) and notifies the master unit that the client flow has ended (815). When the master unit is notified of the end of the client flow from the first network interface unit, AFHT deletes the information of the client flow (816) and returns the result.

In the same procedure, when the third network interface unit recognizes that the server flow has ended (817), it deletes the server flow information in its LFHT (818) and notifies the master unit that the server flow has ended (819). When the master unit is notified of the end of the server flow from the third network interface unit, AFHT deletes the information of the server flow (820) and returns the result.

As a result, in the AFHT of the master unit at the request of the first network interface unit and the third network interface unit, all information about the asymmetric flow is deleted.

When comparing the connection-oriented and non-connection-oriented embodiments, in the case of the connection-oriented embodiments, since there is an explicit flow generation procedure by the SYN / SYN-ACK / ACK message, the network interface units process themselves relatively early. You can check whether the flow is an asymmetric flow or a symmetric flow. In addition, in the case of the connection-oriented embodiment, since there is an explicit flow release procedure by the FIN message, the network interface units can check the release of the flow that they process relatively early to inform the master unit.

On the other hand, in the case of the non-connection-oriented embodiment, since there is no explicit flow generation procedure, each network interface unit is applicable when the opposite flow does not flow to itself until a predetermined number of packets are processed or a predetermined timer value is completed. The flow determines the asymmetric flow.

However, even in the case of a non-connection-oriented protocol, as in the case of a connection-oriented protocol, NIC1 and NIC3 can be configured to maintain traffic information for flows that are not asymmetric flows only and do not provide to the master unit. According to the above-described method, NIC1 and NIC3 can distinguish whether the flow passing through them is a symmetrical flow or a flow belonging to an asymmetrical flow, so the traffic information for the symmetrical flow is stored only by itself and may not be provided to the master unit. . This can have the effect of reducing the communication bandwidth burden between the master unit and the network interface units. Meanwhile, the master unit comprehensively manages information on asymmetric flows of the network redundancy device to which it belongs. The master unit may provide asymmetric flow information managed by an external (user / administrator) request to enable integrated management.

Although described above with reference to preferred embodiments of the present invention, those skilled in the art variously modify and change the present invention without departing from the spirit and scope of the present invention as set forth in the claims below. You can understand that you can.

200: network redundancy device
210, 220, 230, 240: network interface unit
250: master unit
300: network interface unit
310: control interface unit 320: control unit
330: LFHT storage unit 340: flow traffic information storage unit
350: network interface unit
400: master unit
410: control interface unit 420: CPU
430: AFHT storage
440: asymmetric flow traffic information storage unit

Claims (18)

A network redundancy device having a master unit and a plurality of network interface units,
If the information of the flow introduced to itself is not generated, information of the flow is stored, and when the flow corresponds to an asymmetric flow, information of the flow is provided to the master unit, and the flow A plurality of network interface units requesting and receiving information of opposite flows from the master unit; And
Stores information of the flow provided from the network interface unit, checks whether information of the opposite flow for the flow is stored, and if information of the opposite flow is stored, the information of the opposite flow is stored in the network interface unit Includes the master unit provided,
The plurality of network interface units are composed of network interface cards mounted in a slot of one board, and the master unit includes a CPU (Central Processing Unit),
The network redundancy device operates based on a connection-oriented protocol, and when the SYN-ACK packet flows in the network interface unit, information of a flow to which the SYN packet corresponding to the SYN-ACK packet belongs is stored in the network interface unit. If not, the network redundancy device characterized in that the flow is recognized as corresponding to the asymmetric flow. delete delete The method according to claim 1,
When the ACK packet for the SYN-ACK packet flows in, the network interface unit recognizes the flow as an asymmetric flow if information on the flow to which the SYN-ACK packet belongs is not stored in the network interface unit. Redundancy device. A network redundancy device having a master unit and a plurality of network interface units,
If the information of the flow introduced to itself is not generated, information of the flow is stored, and when the flow corresponds to an asymmetric flow, information of the flow is provided to the master unit, and the flow A plurality of network interface units requesting and receiving information of opposite flows from the master unit; And
Stores information of the flow provided from the network interface unit, checks whether information of the opposite flow for the flow is stored, and if information of the opposite flow is stored, the information of the opposite flow is stored in the network interface unit Includes the master unit provided,
The plurality of network interface units are composed of network interface cards mounted in a slot of one board, and the master unit includes a CPU (Central Processing Unit),
The network redundancy device is a network redundancy device that operates based on a connectionless oriented protocol. The method according to claim 5,
The network interface unit recognizes the flow as an asymmetric flow if the flow opposite to the flow does not flow into the network interface unit for a predetermined period. The method according to claim 1,
The network interface unit
A network redundancy device that provides traffic information of flows corresponding to the asymmetric flow to the master unit at the time of traffic processing of the flows corresponding to the asymmetric flow or at a predetermined cycle. delete The method according to claim 1,
The network interface unit
A control interface unit interfacing with the master unit;
A network interface unit that transmits and receives the flow to and from the outside;
A local flow header table storage unit that stores information on flows processed by the network interface unit;
A traffic information storage unit for storing traffic information of the flows; And
When the information flowing through the network interface unit is analyzed and the flow information corresponding to the packet is not generated, the flow information is generated and stored in the local flow header table storage unit, and the flow corresponds to an asymmetric flow. If it does, the information of the flow is provided to the master unit through the control interface unit, and information of the flow opposite to the flow is requested and received from the master unit, and traffic information of the flow is stored in the traffic information storage unit Network redundancy device including control unit. The method according to claim 1,
The master unit
A control interface unit interfacing with the network interface unit;
An asymmetric flow header table storage unit for storing information of an asymmetric flow received from the network interface unit;
A traffic information storage unit for storing traffic information of the asymmetric flows; And
When the information of the opposite flow for the flow is stored by referring to the information of the asymmetric flow received from the network interface unit through the control interface, when the information of the opposite flow is stored, the network interface unit A network redundancy device including a CPU that executes a program that provides information of the opposite flow and stores a function of the flow. A method of operating a network redundancy device having a plurality of network interface units composed of a master unit including a CPU (Central Processing Unit) and network interface cards mounted in a slot of one board,
Analyzing the flow of the network interface unit into the network interface unit;
Storing information of the flow when the network interface unit does not generate the flow information;
The network interface unit determines whether the flow corresponds to an asymmetric flow, and if the flow is an asymmetric flow, provides information on the flow to the master unit, and requests information on the opposite flow for the flow to the master unit to receive To do;
The master unit storing information of the flow provided from the network interface unit; And
The master unit checks whether the information of the opposite flow for the flow is stored, and when the information of the opposite flow is stored, providing the information of the opposite flow to the network interface unit.
The flow is a flow according to a connection-oriented protocol, and if the SYN-ACK packet flows in the network interface unit, if the information of the flow to which the SYN packet corresponding to the SYN-ACK packet belongs is not stored in the network interface unit A network redundancy method, characterized in that the flow is recognized as corresponding to an asymmetric flow. delete delete The method according to claim 11,
When the ACK packet for the SYN-ACK packet flows in, the network interface unit recognizes the flow as an asymmetric flow if information on the flow to which the SYN-ACK packet belongs is not stored in the network interface unit. Redundancy method. A method of operating a network redundancy device having a plurality of network interface units composed of a master unit including a CPU (Central Processing Unit) and network interface cards mounted in a slot of one board,
Analyzing the flow of the network interface unit into the network interface unit;
Storing information of the flow when the network interface unit does not generate the flow information;
The network interface unit determines whether the flow corresponds to an asymmetric flow, and when the flow is an asymmetric flow, provides information on the flow to the master unit, and requests information on the opposite flow for the flow to the master unit to receive To do;
The master unit storing information of the flow provided from the network interface unit; And
And the master unit confirming whether information of the opposite flow for the flow is stored, and providing information of the opposite flow to the network interface unit when the information of the opposite flow is stored.
The flow is a network redundancy method that is a flow according to a non-connection oriented protocol. The method according to claim 15,
The network interface unit recognizes that the flow corresponds to an asymmetric flow if the flow opposite to the flow does not flow into the network interface unit for a predetermined period. The method according to claim 11,
And providing, by the network interface unit, traffic information of flows corresponding to the asymmetric flow to the master unit at a time of processing traffic of flows corresponding to the asymmetric flow or at a predetermined cycle. The method according to claim 17,
And the master unit storing traffic information of flows corresponding to the asymmetric flow provided from the network interface unit.

Images