KR101029416B1 - Ranking data system, ranking query system and ranking computation method for computing large scale ranking in real time - Google Patents

Abstract

A ranking data system, a ranking query system, and a ranking estimation method for estimating a large capacity ranking in real time are disclosed. The ranking data system reads the attribute information of the ranking to be serviced from the ranking scheme table and generates a data structure corresponding to the attribute information in the memory. The ranking is divided according to the score intervals and distributed in a plurality of ranking data systems. A ranking information loading unit which loads ranking information of a corresponding score section among information from a database and loads it into a memory through a data structure generated, and requests a query from a ranking query system that receives a query from an application among a plurality of ranking query systems. And a request processing unit which processes the received request information and the operation according to the received request on the loaded ranking information, and transmits the processed result to the ranking query system.

Ranking, Game, Database, Real Time, Routing Table

Description

RANKING DATA SYSTEM, RANKING QUERY SYSTEM AND RANKING COMPUTATION METHOD FOR COMPUTING LARGE SCALE RANKING IN REAL TIME}

Embodiments according to the present invention relate to a ranking data system, a ranking query system and a ranking calculation method for estimating a large capacity ranking in real time.

The ranking system according to the prior art is generally implemented in two ways as follows.

First, there is a way to use a ranking system that depends on the database. The ranking system stores a numerical value for calculating a ranking using a general DBMS, and calculates and stores the ranking through a database query operation at a predetermined time. The ranking system can calculate the ranking only by the basic database operation, and it is the easiest way to implement because the application only needs to read the ranking information stored in the database.

However, when using a ranking system that depends on the database, it is difficult to calculate the ranking in real time due to the computational load consumed in the database. Therefore, after storing a numerical value for calculating the ranking in the ranking system, the ranking is aggregated every predetermined time (for example, once a day), and the ranking information collected by the application is read. It is not impossible to calculate in real time, but doing so would reduce the scope of the ranking aggregates so that it would not overload the database. After all, the database is not suitable for aggregating real-time rankings because of its relatively slow reactivity.

Second, there is a way to build a ranking server yourself. This method manages the ranking information by implementing the ranking server directly instead of the expensive and responsive DBMS. In this case, the application may exchange a request and a response through the interface defined with the ranking server. In this case, the ranking server conceptually operates as a kind of database cache for the ranking information. Such a ranking server needs to permanently store ranking information managed internally. At this time, the ranking information may be managed as a data file directly by using a separate DBMS inside / outside, or using the file system of the operating system.

However, since all of the ranking information must be processed by one single server, when the traffic for the ranking request is very large or the number of stored ranking information is very large, physical performance limitations are encountered. Although the ranking servers can be distributed and processed by service domains, there is a disadvantage in that the application must know all the information about the ranking servers assigned by service domains. In addition, if a performance problem is encountered in a single service domain as well, it cannot be distributed. In addition, as the ranking server development takes time, it is not easy to apply the ranking to new services. Because the schema for the ranking itself is hard-coded, additional development is often required to change the schema or apply a new ranking.

In the present specification, a ranking data system, a ranking query system, and a ranking calculation method capable of calculating such a large-capacity ranking in real time and efficiently are proposed.

There is provided a ranking data system and a ranking calculation method capable of calculating a large amount of ranking in real time by distributing and loading ranking information in a plurality of ranking data systems and processing the ranking information only for its own score section.

By using the ranking route table, the query is relayed to the ranking data system that matches the query of the application, and the ranking is determined through the counters of the ranking data system's responses and the plurality of ranking data systems stored in the ranking route table. A ranking query system and a ranking estimation method are provided that enable ranking estimation.

A ranking data system, a ranking query system, which can dynamically distribute loads by moving some of the ranking information of the ranking data system with a high load rate to the ranking data system with a low load rate based on the burst rate of the ranking data system; A ranking calculation method is provided.

The ranking data system according to an embodiment of the present invention reads attribute information of a ranking to be serviced from a ranking scheme table, and generates a data structure corresponding to the attribute information in a memory. The ranking information loading unit which receives the ranking information of the corresponding score section among the ranking information distributed in the ranking data system and loads it into the memory through the data structure generated, and the ranking which receives the query from the application among the plurality of ranking query systems. A request receiving unit for receiving a request for a query from the query system, and a request processing unit for processing the operation according to the received request for the loaded ranking information, and transmits the processing result to the ranking query system.

According to an aspect of the invention, the ranking data system may include a load rate calculation unit for calculating its own load rate (burst rate), the load information and the ranking information of the adjacent score interval of the plurality of ranking data system is loaded Some of the jacked ranking information may be moved to the other ranking data system or a part of the ranking information loaded in the other ranking data system based on the load rate of the other ranking data system.

According to an aspect of the present invention, some of the loaded ranking information may be moved to another ranking data system when the ratio of the load ratio of the other ranking data system of its load rate is more than a certain ratio.

According to an aspect of the present invention, a plurality of ranking data systems may be loaded so that some of the score sections overlap each other for each ranking data system that loads the ranking information of the adjacent score sections, and some of the ranking information moved may be different rankings. The data system may include ranking information overlapping each other.

According to one aspect of the invention, the ranking data system may further include a load rate transmission unit for broadcasting the load rate to a plurality of ranking query system, the load rate may be updated in the ranking routing table included in the plurality of ranking query system .

According to an aspect of the present invention, the score section may be changed based on the load ratio in the plurality of ranking query systems, and some of the ranking information moved may include ranking information determined according to the change of the score section. .

According to an aspect of the present invention, the attribute information may be generated through a ranking management tool provided to the administrator.

According to an aspect of the present invention, the load ratio may be determined according to the throughput per second (TPS) and the free memory capacity.

According to an aspect of the invention, the ranking data system may further include a backup unit for periodically backing up its ranking information to the database.

According to an aspect of the present invention, a query may be transmitted to a ranking data system determined from a plurality of ranking data systems based on the ranking routing table in the ranking query system.

According to an aspect of the present invention, the ranking may be calculated in the ranking query system based on the results processed by the request processing unit and the counts for each of the plurality of ranking data systems stored in the ranking routing table, and the count may request ranking information or It may include the number of ranking service using the subject to register the ranking information.

In a ranking calculation method according to an embodiment of the present invention, a data structure generation unit of a ranking data system reads attribute information of a ranking to be serviced from a ranking scheme table, and generates a data structure corresponding to the attribute information in a memory. Receiving the ranking information of a corresponding score section among the ranking information divided by the score section in the ranking information loading unit of the system and distributed to the plurality of ranking data systems, and loading the memory into the memory through a data structure generated; Receiving a request for a query from the ranking query system that received a query from the application among the plurality of ranking query system in the request receiving unit of the ranking data system and the operation according to the request received in the request processing unit of the ranking data system For and process the results Transmitting a query King system.

The ranking query system according to an embodiment of the present invention receives at least one of attribute information and a load rate from a plurality of ranking data systems to create or update a ranking routing table, analyzes a query received from an application, and rank routing A query processor for transmitting a query to a ranking data system including ranking information corresponding to a query among a plurality of ranking data systems, a response processor for receiving a response of the ranking data system for the query and transmitting the response to an application; And a score section updating unit for updating the score section of the ranking information included in the ranking data system to the ranking routing table based on the load ratio.

The ranking calculation system according to an embodiment of the present invention comprises generating or updating a ranking routing table by receiving at least one of attribute information and a load rate from a plurality of ranking data systems in a table management unit of a ranking query system, and querying a ranking query system. Analyzing the query received from the application in the processing unit and transmitting the query to a ranking data system including ranking information corresponding to the query among the plurality of ranking data systems based on the ranking routing table, in the response processing unit of the ranking query system Receiving a response of the ranking data system for the query and transmitting the response to the application and the score interval updating unit of the ranking query system updates the score interval of the ranking information each of the plurality of ranking data system in the ranking routing table based on the load ratio Ha Score section includes updating unit.

A large capacity ranking can be calculated in real time by distributing and loading ranking information in a plurality of ranking data systems and processing the ranking information only for its own score section.

By using the ranking route table, the query is relayed to the ranking data system that matches the query of the application, and the ranking is determined through the counters of the ranking data system's responses and the plurality of ranking data systems stored in the ranking route table. The ranking can be calculated.

Based on the load rate of the ranking data system, by moving some of the ranking information of the ranking data system with a high load rate to the ranking data system with a low load rate, the load can be distributed dynamically.

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

1 is a view showing an overview of an overall system for providing a ranking service according to an embodiment of the present invention.

Here, the ranking service system 100 may refer to a server group configured of a plurality of ranking query servers and a plurality of ranking data servers to provide a real-time ranking service. At this time, the ranking information is internally distributed in order to provide a large amount of ranking information in real time at a high speed. At this time, the user of the ranking service can use the ranking service using an interface provided as a service without knowing the internal structure.

The ranking management tool 110 is a tool designed to manage the ranking master scheme 120. For example, the ranking management tool 110 may be used to separate a development task from a ranking service operation task. The ranking management tool 110 may provide general functions necessary for operation such as information on the ranking master scheme 120 as well as the process of deleting the illegal user ranking and checking the ranking log.

The ranking master scheme 120 sets meta information about the ranking, such as an attribute of the ranking, and may be maintained as a ranking scheme table. In this case, the ranking data server included in the ranking service system 100 may generate a data structure corresponding to the ranking to be serviced through the ranking scheme table. At this time, the ranking scheme table includes the ranking service domain name, the ranking period type (daily, weekly, monthly or cumulative), the ranking sort attribute (ascending or descending), the ranking tie processing method (first registrant first or latest registrant first), and the number of ranking types. Various meta information such as designation may be set.

The database 130 is an external storage for backing up the ranking information stored in the memory of the ranking data server. The first ranking data server may generate the data structure and receive the necessary ranking information from the database and load it into the memory.

The application 140 may refer to a ranking service user that requests ranking information or registers the ranking information. For example, it may be any program that runs on a terminal used by a user.

2 is a view showing an overview of a ranking service system according to an embodiment of the present invention. As briefly described with reference to FIG. 1, the ranking service system 100 may include a plurality of ranking data systems 210 and a plurality of ranking query systems 220. In this case, the plurality of ranking data systems 210 may correspond to the plurality of ranking data servers described with reference to FIG. 1, and the plurality of ranking query systems 220 may correspond to the plurality of ranking query servers described with reference to FIG. 1.

At this time, when the application 140 requests the ranking information to the ranking service system 100, it actually performs an anycast operation with one of the plurality of ranking query systems 220. In this case, anycast may be performed through the Layer-4 (L4) switch or may be performed in software in the application 140.

At this time, the ranking query system for receiving a query according to the request of the ranking information among the plurality of ranking query system 220 analyzes the received query, the ranking data system including the corresponding ranking information among the plurality of ranking data system 210 In response to the request, the query may be delivered, the response may be received from the ranking data system, and then transferred back to the application 140. In this case, since the plurality of ranking query system 220 needs to distribute a large amount of traffic, the ranking query system 220 may be flexibly expanded laterally.

The ranking data system 210 stores actual ranking information, and stores the ranking information in a memory for fast processing. The plurality of ranking data systems 210 may generate an optimal data structure by using attribute information of the ranking included in the ranking scheme table. In this case, since the physical memory is also limited, the plurality of ranking data systems 210 may also be horizontally extended.

3 is a block diagram illustrating an internal configuration of a ranking data system according to an embodiment of the present invention. The ranking data system 300 according to the present embodiment may be one of the plurality of ranking data systems 210 described with reference to FIG. 2. In addition, as shown in FIG. 3, the ranking data system 300 includes a data structure generator 310, a ranking information loader 320, a request receiver 330, a request processor 340, and a load factor calculator 350. ), The load factor transmitter 360 and the backup unit 370 may be included.

The data structure generation unit 310 reads the attribute information of the ranking to be serviced from the ranking scheme table, and generates a data structure corresponding to the attribute information in the memory. At this time, the attribute information may be generated through a ranking management tool provided to the administrator.

The ranking information loading unit 320 receives the ranking information of the corresponding score section among the ranking information divided by the score section distributed in a plurality of ranking data systems from the database and loads the ranking information into the memory through the data structure generated. In more detail, the ranking may be divided by service domain, and again, the ranking information may be divided according to a score section. At this time, the ranking data system 300 may calculate the ranking only for the score section assigned to it, for this purpose, the ranking information loading unit 320 receives the ranking information of the corresponding score section from the database and loads it into the memory. Can be.

The request receiving unit 330 receives a request for a query from a ranking query system that receives a query from an application among a plurality of ranking query systems. In this case, the query may be transmitted to one of the plurality of ranking data systems based on the ranking routing table in the ranking query system. This ranking routing table may have the same value in all of the plurality of ranking query systems. For example, when the ranking information loaded is updated, the ranking data system 300 may broadcast the information to a plurality of ranking query systems so that the ranking routing table may be updated in the same manner.

The request processor 340 processes an operation according to the received request with respect to the loaded ranking information, and transmits the processed result to the ranking query system. The ranking may be calculated in the ranking query system based on the results processed by the request processing unit 340 and the counts for each of the plurality of ranking data systems stored in the ranking routing table, and the count may be used to request ranking information or register ranking information. It may include the number of users using the ranking service. Therefore, since the ranking query system can know the number of users included in the score interval corresponding to each of the plurality of ranking data systems through the count stored in the ranking routing table, knowing the ranking in one of the plurality of ranking data systems, The ranking of the application requesting the query can be calculated in the entire ranking service subject. This ranking calculation will be described in more detail with reference to FIG. 5 below.

The load rate calculation unit 350 calculates its own burst rate. Here, the load rate may be determined according to the throughput per second (TPS) and the free memory capacity. For example, the load factor may be calculated as in Equation 1 below.

Load factor = Exceed TPS * 100 + (1,000,000 free memory capacity)

Here, the 'Exceed TPS' may have a value of 'Current TPS-5000', and when 'Exceed TPS' is negative, it may be corrected to '0'. In addition, the unit of the free memory capacity may be megabytes.

At this time, part of the ranking information jacked based on its load rate and the load rate of another ranking data system in which ranking information of adjacent score sections among the plurality of ranking data systems is loaded is moved to another ranking data system or loaded in another ranking data system. A part of the ranked ranking information may be moved. In more detail, some of the loaded ranking information may be moved to another ranking data system when the ratio of its load rate to the load rate of another ranking data system is more than a certain ratio. For example, in the ranking data systems H and I where the score intervals are adjacent to each other, if the load ratio of H is 1.5 times larger than the load ratio of I, some of the ranking information loaded in H may be moved to I, thereby reducing the load of H. have. In this case, I may also compare the load ratio with the ranking data system J adjacent to the score section, and load may be dynamically distributed by sequentially comparing the load rates of all the ranking data systems with adjacent score sections.

In addition, for each of the ranking data systems that load the ranking information of the score sections adjacent to each other in the plurality of ranking data systems, some of the score sections may be stacked to overlap each other, and at this time, some of the ranking information may be overlapped with other ranking data systems. The ranking information may be included. When requesting the adjacent ranking information for the ranking information located at the end of the score interval, the score interval is set to overlap each other to prevent the reading request to the two ranking data server, so that the read request for the list is This is to limit the occurrence to only one ranking data server. In this case, the overlapping ranking information may be moved according to the load ratio.

The load rate transmitter 360 broadcasts the load rate to a plurality of ranking query systems. In this case, the load ratio may be updated in a ranking routing table included in a plurality of ranking query systems. In this case, the score section may be changed based on the load ratio in the plurality of ranking query systems, and the some ranking information moved may include ranking information determined according to the change of the score section. That is, the movement of the ranking information according to the load ratio may be performed through data movement between the ranking data systems, but may also be achieved by receiving and loading the ranking information of the score section changed in the ranking routing table from the database in each ranking data system.

The backup unit 370 periodically backs up its ranking information to the database.

4 is a block diagram illustrating an internal configuration of a ranking query system according to an embodiment of the present invention. The ranking query system 400 according to the present embodiment may be one of the plurality of ranking query systems 220 described with reference to FIG. 2. In addition, the ranking query system 400 may include a table manager 410, a query processor 420, a response processor 430, and a score interval update unit 440 as shown in FIG. 4.

The table manager 410 receives at least one of the attribute information and the load factor from the plurality of ranking data systems to generate or update the ranking routing table. In this case, the load ratio may be determined and broadcasted according to throughput per second (TPS) and free memory capacity in the plurality of ranking data systems, respectively. In addition, each of the plurality of ranking data systems may read attribute information of ranking to be serviced from the ranking scheme table, and generate a data structure corresponding to the attribute information in the memory. That is, the table manager 410 may generate or update the ranking routing table by receiving attribute information and load rates respectively received from the plurality of ranking data systems.

In this case, each of the plurality of ranking data systems may be divided according to the score section and loaded into the memory through the data structure generated by receiving the ranking information of the corresponding score section among the ranking information distributed in the plurality of ranking data systems from the database. have. In this case, each of the ranking data systems for loading the ranking information of the score section adjacent to each other among the plurality of ranking data system may be stacked so that a part of the score section is overlapped with each other, the ranking routing table is the score section of the ranking information loaded without overlapping Denotes a read interval, and a score interval of non-overlapping ranking information and a score interval of overlapped stacked ranking information are stored as right intervals, respectively. That is, the write operation may occur in two ranking data systems at the same time, but the read operation may be limited to occur in one ranking data system.

The query processor 420 analyzes a query received from an application and transmits the query to a ranking data system including ranking information corresponding to the query among the plurality of ranking data systems based on the ranking routing table. In this case, the query from the application may be transmitted to one of a plurality of ranking query systems through an anycast transmission scheme.

The response processor 430 receives a response of the ranking data system for the query and transmits the response to the application. In this case, the ranking routing table may include a count, which is the number of ranking service users who request ranking information or register ranking information, for each of the plurality of ranking data systems, and the ranking may be calculated based on the response and the count.

For example, the total score section is from 0 to 99 points, and is divided into 10 score sections in units of 10 points, and corresponding score sections in the 10 ranking data systems of the first ranking data system to the tenth ranking data system, respectively. Suppose we include ranking information for. At this time, if the query requests the ranking included in the score interval of 70 to 79 points, the ranking query system 400 relays the query to the eighth ranking data system to receive a response within the score interval of 70 to 79 points You can get a rank at. In addition, the ranking query system 400 checks the counts of the ninth ranking data system and the tenth ranking data system in the ranking routing table, and is included in the score section of 80 to 89 points and the score section of 90 to 99 points. The number of subjects using the ranking service may be checked to obtain a ranking outside the desired score section. That is, the ranking query system 400 may simply calculate the ranking of the application by summing the counts of the ninth ranking data system and the tenth ranking data system and rankings within a score section of 70 to 79 points.

The score section updater 440 updates the score section of the ranking information included in each of the ranking data systems in the ranking routing table based on the load ratio. In this case, when the ratio between the load ratios of the ranking data systems adjacent to each other is equal to or greater than A, the score interval update unit 440 may include the score intervals of the ranking information included in the ranking data systems adjacent to each other based on the load ratios in the ranking routing table. Can be updated. Here, the ranking data systems adjacent to each other may mean a ranking data system including ranking information of score sections adjacent to each other.

For example, in the ranking data systems H and I where the score intervals are adjacent to each other, if the load ratio of H is 1.5 times larger than the load ratio of I, some of the ranking information loaded in H may be moved to I, thereby reducing the load of H. have. In this case, I may also compare the load ratio with the ranking data system J adjacent to the score section, and load may be dynamically distributed by sequentially comparing the load rates of all the ranking data systems with adjacent score sections.

As such, when using the ranking query system and the ranking data system according to the embodiments of the present invention, the ranking information is distributed and loaded in the plurality of ranking data systems, and the ranking information is processed only for its own score section in real time. A large amount of ranking can be calculated, and the ranking route system relays the query to the ranking data system suitable for the application's query, and the counter of each ranking data system stored in the ranking route system's response and the ranking route system By determining the ranking, a large capacity ranking can be calculated in real time. Further, by moving some of the ranking information of the ranking data system having a high load ratio to the ranking data system having a low load ratio based on the load ratio of the ranking data system, the load can be distributed dynamically.

5 is an example of a ranking routing table included in a ranking query system according to an embodiment of the present invention. The first table 510 includes respective game identifiers and table pointers. That is, the first table 510 primarily shows ranking information classified by service domains. The second table 520 divides the score section into a reading section and a writing section for one game, so that the minimum and maximum values for each score section, the count representing the number of ranking service users included in the score section, and the corresponding score section It includes the address of the ranking data system loaded with the ranking information of and the load ratio of each ranking data system.

Suppose that there are only three ranking data systems shown in the second table 520, and the total score interval is '0' to '10000'. In this case, if the ranking query system receives the ranking '100' as a response from the second ranking data system having the read interval between '4000' and '6000', the ranking of the application is' at the read interval '7000'. It can be seen that the sum '734' of the third ranking data system between 9000 'points and' 834 'is the sum of the responses' 100'.

In addition, since the load rate of the third ranking data system is '7832' and is 1.5 times larger than the load rate of the second ranking data system '4423', some of the ranking information of the third ranking data system is transmitted to the second ranking data system. The load of the third ranking data system can be reduced. In this case, the third ranking data system and the second ranking data system have ranking information corresponding to a score section between '6000' points and '7000' points through the write section minimum value and the write section maximum value of the second table 520. It can be seen that they are stacked on top of each other. That is, the ranking query system may update the maximum value of the read interval of the second ranking data system from the existing '6000' point to '7000' point, and update the maximum value of the write interval from '7000' point to '8000' point. Can be. In addition, the ranking query system may update the minimum value of the read interval of the third ranking data system from '7000' to '8000' and the minimum value of the write interval from '6000' to '7000'. That is, if the application requests the ranking of the user having a score of '7500' before updating, the ranking query system relays the request to the third ranking data system. However, after the update, the request is relayed to the second ranking data system, thereby reducing the load of the third ranking data system. In this case, the '7500' points do not correspond to the reading intervals of both the second ranking data system and the third ranking data system, but the ranking information of the corresponding scores is overlapped in both the second ranking data system and the third ranking data system. Therefore, the ranking query system may relay the request to the ranking data system having a small load rate according to the load rate. In this case, the update may be applied by directly transmitting the corresponding ranking information from the third ranking data system to the second ranking data system or by reloading the ranking information according to the corresponding score section from the database according to the changed section.

If the load ratio is increased due to the changed score interval in the second ranking data system, the above operation is repeated with the first ranking data system having a read interval of '0 to' 3000 ', thereby spreading the load evenly over the entire service area. Can be.

6 is an example of ranking information loaded into a ranking data system according to an embodiment of the present invention. The first section 601 to the fourth section 604 may mean a score section corresponding to each of the four ranking data systems. In this case, as shown in FIG. 6, each of the ranking data systems may overlap a part of the score intervals. This is to prevent a situation in which a request should be relayed to two ranking data systems when adjacent ranking information is requested for ranking information located at both ends of a score section. For example, if ranking information of '2999' points and ranking information of '3000' points are needed, the ranking information for the '4000' points and the '4001 points' is a ranking data system for the first section 601. If necessary, by relaying the request to the ranking data system for the second interval 602, the read request may be limited to one ranking data server even if a write request occurs in up to two ranking data systems.

7 is a flowchart illustrating a ranking calculation method according to an embodiment of the present invention. The ranking calculation method according to the present embodiment may be performed by the ranking data system 300 described with reference to FIG. 3. In FIG. 3, a ranking calculation method will be described by describing a process in which each step is performed by the ranking data system 300.

In step 710, the ranking data system 300 reads the attribute information of the ranking to be serviced from the ranking scheme table, and creates a data structure corresponding to the attribute information in memory. At this time, the attribute information may be generated through a ranking management tool provided to the administrator.

In step 720, the ranking data system 300 is divided according to the score intervals and distributed to the plurality of ranking data systems among the ranking information of the corresponding score intervals among the ranking information received from the database to the memory structure through the data structure generated Load it. In more detail, the ranking may be divided by service domain, and again, the ranking information may be divided according to a score section. In this case, the ranking data system 300 may calculate the ranking only for the score section allocated to the ranking data system 300, and receive the ranking information of the corresponding score section from the database and load the ranking information in the memory.

In operation 730, the ranking data system 300 receives a request for a query from a ranking query system that receives a query from an application among a plurality of ranking query systems. In this case, the query may be transmitted to one of the plurality of ranking data systems based on the ranking routing table in the ranking query system. This ranking routing table may have the same value in all of the plurality of ranking query systems. For example, when the ranking information loaded is updated, the ranking data system 300 may broadcast the information to a plurality of ranking query systems so that the ranking routing table may be updated in the same manner.

In operation 740, the ranking data system 300 processes an operation according to the received request with respect to the loaded ranking information, and transmits the processed result to the ranking query system. The ranking may be calculated in the ranking query system based on the results processed by the ranking data system 300 and the counts for each of the plurality of ranking data systems stored in the ranking routing table, the counts requesting the ranking information or registering the ranking information. It may include the number of users using the ranking service. Therefore, since the ranking query system can know the number of users included in the score interval corresponding to each of the plurality of ranking data systems through the count stored in the ranking routing table, knowing the ranking in one of the plurality of ranking data systems, The ranking of the application requesting the query can be calculated in the entire ranking service subject. This ranking calculation will be described in more detail with reference to FIG. 5 below.

In step 750, ranking data system 300 calculates its burst rate. Here, the load ratio may be determined according to the throughput per second (TPS) and the free memory capacity.

At this time, part of the ranking information jacked based on its load rate and the load rate of another ranking data system in which ranking information of adjacent score sections among the plurality of ranking data systems is loaded is moved to another ranking data system or loaded in another ranking data system. A part of the ranked ranking information may be moved. In more detail, some of the loaded ranking information may be moved to another ranking data system when the ratio of its load rate to the load rate of another ranking data system is more than a certain ratio. For example, in the ranking data systems H and I where the score intervals are adjacent to each other, if the load ratio of H is 1.5 times larger than the load ratio of I, some of the ranking information loaded in H may be moved to I, thereby reducing the load of H. have. In this case, I may also load balance dynamically by comparing the load rate with the ranking data system J adjacent to the score section and sequentially comparing the load rates of all the ranking data systems with adjacent score sections.

In addition, for each of the ranking data systems that load the ranking information of the score sections adjacent to each other in the plurality of ranking data systems, some of the score sections may be stacked to overlap each other, and at this time, some of the ranking information may be overlapped with other ranking data systems. The ranking information may be included. When requesting the adjacent ranking information for the ranking information located at the end of the score interval, the score interval is set to overlap each other to prevent the reading request to the two ranking data server, so that the read request for the list is This is to limit the occurrence to only one ranking data server. In this case, the overlapping ranking information may be moved according to the load ratio.

In step 760, ranking data system 300 broadcasts the load rate to a plurality of ranking query systems. In this case, the load ratio may be updated in a ranking routing table included in a plurality of ranking query systems. In this case, the score section may be changed based on the load ratio in the plurality of ranking query systems, and the some ranking information moved may include ranking information determined according to the change of the score section. That is, the movement of the ranking information according to the load ratio may be performed through data movement between the ranking data systems, but may also be achieved by receiving and loading the ranking information of the score section changed in the ranking routing table from the database in each ranking data system.

In step 770, ranking data system 300 periodically backs up its ranking information to the database.

The above-described steps 710 to 770 may include a data structure generation unit, a ranking information loading unit, a request receiving unit, a request processing unit, a load rate calculating unit, a load rate transmitting unit, and a backup unit that the ranking data system 300 may include. Each can be performed.

8 is a flowchart illustrating a ranking calculation method according to another embodiment of the present invention. The ranking calculation method according to the present embodiment may be performed through the ranking query system 400 described with reference to FIG. 4. In FIG. 8, a ranking calculation method will be described by describing a process of performing each step through the ranking query system 400.

In step 810, the ranking query system 400 receives at least one of the attribute information and the load factor from the plurality of ranking data system to generate or update the ranking routing table. In this case, the load ratio may be determined and broadcasted according to the TPS and the free memory capacity in the plurality of ranking data systems, respectively. In addition, each of the plurality of ranking data systems may read attribute information of ranking to be serviced from the ranking scheme table, and generate a data structure corresponding to the attribute information in the memory. That is, the ranking query system 400 may generate or update the ranking routing table by receiving attribute information and load rates respectively received from the plurality of ranking data systems.

In this case, each of the plurality of ranking data systems may be divided according to the score section and loaded into the memory through the data structure generated by receiving the ranking information of the corresponding score section among the ranking information distributed in the plurality of ranking data systems from the database. have. In this case, each of the ranking data systems for loading the ranking information of the score section adjacent to each other among the plurality of ranking data system may be stacked so that a part of the score section is overlapped with each other, the ranking routing table is the score section of the ranking information loaded without overlapping As a reading section, the score section of the non-overlapping ranking information and the score section of the overlapping ranking information may be stored as write sections, respectively. That is, the write operation may occur in two ranking data systems at the same time, but the read operation may be limited to occur in one ranking data system.

In operation 810, the ranking query system 400 analyzes a query received from an application, and includes a ranking data system including ranking information corresponding to the query among the plurality of ranking data systems based on the ranking routing table. Send the query. In this case, the query from the application may be transmitted to one of a plurality of ranking query systems through an anycast transmission scheme.

In step 810, ranking query system 400 receives the response of the ranking data system for the query and transmits to the application. In this case, the ranking routing table may include a count, which is the number of ranking service users who request ranking information or register ranking information, for each of the plurality of ranking data systems, and the ranking may be calculated based on the response and the count.

For example, the total score section is from 0 to 99 points, and is divided into 10 score sections in units of 10 points, and corresponding score sections in the 10 ranking data systems of the first ranking data system to the tenth ranking data system, respectively. Suppose we include ranking information for. At this time, if the query requests the ranking included in the score interval of 70 to 79 points, the ranking query system 400 relays the query to the eighth ranking data system to receive a response within the score interval of 70 to 79 points You can get a rank at. In addition, the ranking query system 400 checks the counts of the ninth ranking data system and the tenth ranking data system in the ranking routing table, and includes them in the score section of 80 to 89 points and the score section of 90 to 99 points. By checking the number of users using the ranking service, the ranking can be obtained outside the desired score section. That is, the ranking query system 400 may simply calculate the ranking of the application by summing the counts of the ninth ranking data system and the tenth ranking data system with the ranks within the score interval of 70 to 79 points.

In step 810, the ranking query system 400 updates the score section of the ranking information included in each of the plurality of ranking data system in the ranking routing table based on the load ratio. In this case, when the ratio between the load ratios of the ranking data systems adjacent to each other is A or more, the ranking query system 400 updates the score intervals of the ranking information included in the ranking data systems adjacent to each other based on the load ratios in the ranking routing table. can do. Here, the ranking data systems adjacent to each other may mean a ranking data system including ranking information of score sections adjacent to each other.

For example, in the ranking data systems H and I where the score intervals are adjacent to each other, if the load ratio of H is 1.5 times larger than the load ratio of I, some of the ranking information loaded in H may be moved to I, thereby reducing the load of H. have. In this case, I may also compare the load ratio with the ranking data system J adjacent to the score section, and load may be dynamically distributed by sequentially comparing the load rates of all the ranking data systems with adjacent score sections.

Here, the above-described steps 810 to 840 may be performed by the table manager, the query processor, the response processor, and the score interval update unit, which the ranking data system 400 may include.

As described above, using the ranking calculation method according to the embodiments of the present invention, the ranking information is distributed and stacked in a plurality of ranking data systems, and the ranking information is processed only for its own score section, thereby enabling a large-scale ranking in real time. By calculating the ranking, relay the query to the ranking data system that matches the query of the application using the ranking route table, and determine the ranking through the counter of each of the plurality of ranking data systems stored in the ranking data table's response and the ranking route table. In addition, large-scale ranking can be calculated in real time. Further, by moving some of the ranking information of the ranking data system having a high load ratio to the ranking data system having a low load ratio based on the load ratio of the ranking data system, the load can be distributed dynamically.

9 is an example for explaining the basic operation of the ranking data system and the ranking query system according to an embodiment of the present invention.

In operation 910, the attribute information to be generated may be generated through the attribute of the ranking input through the ranking management tool, and the attribute information of the ranking may be registered in the ranking scheme table. At this time, when the ranking service is started in step 920, the ranking data system reads the ranking information to be served from the ranking scheme table, and creates a data structure corresponding thereto in main memory, and generates ranking information backed up from the database. Can be loaded into internal data structures.

In step 930, the ranking query system may read the attribute information generated in each ranking data system and generate a ranking routing table. In step 940, when the ranking query system receives a query for a ranking request from an application, the ranking request is queried for a ranking routing table in which ranking information corresponding information corresponds to the ranking request, and relays the request to the corresponding ranking data system. Can be. In this case, the query for the ranking request from the application may be received through an anycast transmission scheme for the plurality of ranking query systems.

In operation 950, the ranking data system searches for an internal data structure in which the ranking information is stored, performs a required operation, and transfers the result to the requested ranking query system. In operation 960, the ranking query system may deliver the result received from the ranking data system to the application that has transmitted the query. In this case, as described above, the overall ranking may be calculated by using the ranking routing table and the received result by the ranking query system and then transferred to the application.

In operation 970, the result received by the application may be transmitted to the user, and in operation 980, the ranking data system may periodically back up its ranking information to the database.

10 is an example for explaining an operation of dynamically distributing a load in a ranking data system and a ranking query system according to an embodiment of the present invention.

In step S1010, the ranking data system periodically broadcasts its load rate to the ranking query system. In step 1020, the ranking query system checks whether a ratio between load rates is equal to or greater than A for all ranking data systems including ranking information of score sections adjacent to each other. In this case, step 1020 may be repeatedly performed until the ratio between the load ratios is equal to or greater than A through step 1030.

In step 1040, the area of the adjacent buffer is adjusted by changing the values of the read interval and the write interval of the ranking routing table in the ranking query system. The adjacent buffer is an area for storing ranking information of overlapping score sections in a ranking data system in which ranking information of adjacent score sections is loaded. A low-loading ranking data system stores ranking information of adjacent buffers in a dedicated area. The new adjacent buffer area is set, and the ranking data system having a high load ratio can convert a portion of the dedicated area into the adjacent buffer area. When the operation is performed recursively for a plurality of ranking data systems according to the load rate, the ranking data system with a high load rate narrows the corresponding score section, and the ranking data system with a low load rate widens the corresponding score section to the entire service area. The load can be evenly distributed evenly over time.

Embodiments according to the present invention can be implemented in the form of program instructions that can be executed by various computer means can be recorded on a computer readable medium. The computer readable medium may include program instructions, file data, data structures, etc. alone or in combination. Program instructions recorded on the media may be those specially designed and constructed for the purposes of the present invention, or they may be of the kind well-known and available to those having skill in the computer software arts. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tape, optical media such as CD-ROMs, DVDs, and magnetic disks, such as floppy disks. Magneto-optical media, and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine code generated by a compiler, but also high-level language code that can be executed by a computer using an interpreter or the like. The hardware device described above may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

As described above, the present invention has been described by specific embodiments such as specific components and the like. For those skilled in the art to which the present invention pertains, various modifications and variations are possible.

Therefore, the spirit of the present invention should not be limited to the described embodiments, and all of the equivalents or equivalents of the claims as well as the claims to be described later will belong to the scope of the present invention. .

1 is a view showing an overview of an entire system for providing a ranking service according to an embodiment of the present invention.

2 is a view showing an overview of a ranking service system according to an embodiment of the present invention.

3 is a block diagram illustrating an internal configuration of a ranking data system according to an embodiment of the present invention.

4 is a block diagram illustrating an internal configuration of a ranking query system according to an embodiment of the present invention.

5 is an example of a ranking routing table included in a ranking query system according to an embodiment of the present invention.

6 is an example of ranking information loaded into a ranking data system according to an embodiment of the present invention.

7 is a flowchart illustrating a ranking calculation method according to an embodiment of the present invention.

8 is a flowchart illustrating a ranking calculation method according to another embodiment of the present invention.

9 is an example for explaining the basic operation of the ranking data system and the ranking query system according to an embodiment of the present invention.

10 is an example for explaining an operation of dynamically distributing a load in a ranking data system and a ranking query system according to an embodiment of the present invention.

<Explanation of symbols for the main parts of the drawings>

210: a plurality of ranking data systems

220: a plurality of ranking query system

300: ranking data system

400: Ranking Query System

Claims (19)

A data structure generation unit reading attribute information of a ranking to be serviced from a ranking scheme table and generating a data structure corresponding to the attribute information in a memory; A ranking information stacker configured to receive ranking information of a corresponding score section among the ranking information divided by a score section and distributed in a plurality of ranking data systems from a database and load the ranking information into the memory through the generated data structure; A request receiving unit which receives a request for the query from a ranking query system that receives a query from an application among a plurality of ranking query systems; And Request processing unit for processing the operation according to the received request for the loaded ranking information, and transmits the processing result to the ranking query system Including, ranking data system. The method of claim 1, Load rate calculation unit that calculates its own burst rate Including, Some of the jacked ranking information is moved to the other ranking data system or the other ranking data based on the load rate of the other ranking data system loaded with the own load rate and the ranking information of the adjacent score section among the plurality of ranking data systems. And a portion of the ranking information loaded in the system is moved. The method of claim 2, And some of the loaded ranking information is moved to the other ranking data system when the ratio of the own loading rate to the load rate of the other ranking data system is above a certain ratio. The method of claim 2, A portion of the score intervals are stacked to overlap each other for each ranking data system for loading ranking information of score sections adjacent to each other in the plurality of ranking data systems, The ranking information of the moved portion includes ranking information overlapping with each other ranking data system, ranking data system. The method of claim 2, A load rate transmitter for broadcasting the load rate to a plurality of ranking query systems More, And the load factor is updated in a ranking routing table included in the plurality of ranking query systems. The method of claim 5, The score interval is changed based on the load ratio in the plurality of ranking query system, The ranking information of the moved portion includes ranking information determined according to the change of the score interval, ranking data system. The method of claim 2, Wherein said load rate is determined in accordance with throughput per second (TPS) and free memory capacity. The method of claim 1, The query is sent to one of the plurality of ranking data systems based on a ranking routing table in the ranking query system. The method of claim 1, A ranking is calculated in the ranking query system based on a result processed by the request processing unit and a count for each of the plurality of ranking data systems stored in a ranking routing table. The count includes a number of ranking service usage subjects that request ranking information or register the ranking information. A table manager which receives at least one of attribute information and a load factor from a plurality of ranking data systems to generate or update a ranking routing table; A query processing unit for analyzing a query received from an application and transmitting the query to a ranking data system including ranking information corresponding to the query among the plurality of ranking data systems based on the ranking routing table; And A response processor that receives a response of the ranking data system to the query and transmits the response to the application Ranking query system comprising a. The method of claim 10, A score section updating unit for updating the score section of the ranking information included in the plurality of ranking data systems to the ranking routing table based on the load ratio Ranking query system further comprising. The method of claim 11, The score section update unit, And when the ratio between the load rates of the adjacent ranking data systems is equal to or greater than A, updating the score intervals of ranking information included in the adjacent ranking data systems, respectively, in the ranking routing table based on the load rates. The method of claim 10, And the load ratio is determined and broadcasted according to throughput per second (TPS) and free memory capacity in the plurality of ranking data systems, respectively. The method of claim 10, Some of the plurality of ranking data systems are stacked so that some of the score sections overlap each other for each ranking data system for loading ranking information of adjacent score sections, The ranking routing table reads the score intervals of the non-overlapping ranking information, and reads the score intervals of the non-overlapping ranking information and the score intervals of the redundantly stacked ranking information. Ranking query system for storing the intervals, respectively. The method of claim 10, The query from the application is transmitted to one of a plurality of ranking query system through an anycast transmission scheme. The method of claim 10, The ranking routing table includes a count for each of the plurality of ranking data systems, the count being a number of ranking service users who request ranking information or register ranking information. A ranking query system is calculated based on the response and the count. In the method for calculating the ranking in the ranking data system, Reading, by a data structure generation unit of the ranking data system, attribute information of a ranking to be serviced from a ranking scheme table, and generating a data structure corresponding to the attribute information in a memory; The ranking information loading unit of the ranking data system is divided according to the score section and distributed among the ranking information of the plurality of ranking data system among the ranking information of the corresponding score section received from the database to the memory through the generated data structure Loading; Receiving a request for the query from a ranking query system that receives a query from an application among a plurality of ranking query systems in a request receiving unit of the ranking data system; And A request processing unit of the ranking data system processes an operation according to the received request with respect to the loaded ranking information, and transmits the processed result to the ranking query system Ranking calculation method comprising a. In the method for calculating the ranking in the ranking query system, Generating or updating a ranking routing table by receiving at least one of attribute information and a load factor from a plurality of ranking data systems in a table management unit of the ranking query system; The query processing unit of the ranking query system analyzes the query received from the application and transmits the query to a ranking data system including ranking information corresponding to the query among the plurality of ranking data systems based on the ranking routing table. step; And Receiving, by the response processor of the ranking query system, the response of the ranking data system for the query and transmitting the response to the application; Including, ranking calculation method. A computer-readable recording medium having recorded thereon a program for performing the method of claim 17 or 18.

Images