JP2018007183A - Communication method and load distribution system - Google Patents

Abstract

Provided is a technique for, when establishing a connection between a processing request source and one of a plurality of processing devices, effectively using resources when another processing device communicates with the request source. A first distribution unit that distributes each of the plurality of processing requests received; a plurality of second distribution units that further distribute each of the plurality of processing requests distributed by the first distribution unit; In a communication method in a load distribution system including a plurality of processing devices that respond to each of a plurality of processing requests allocated by any of the second allocation units, the first processing device may include any one of the second allocation units. A first connection is established between the first processing device and the request source of the processing request in response to the first processing request distributed by the unit, and the second processing device responds to an instruction from a source other than the request source. Te, by using a part of the already established first connection, by communicating with the requester, attempt to resolve the above problems. [Selection] Figure 3

Description

  The present specification relates to a communication method and a load distribution system.

  With the spread of network technology, the traffic volume of data transmitted and received on the network has increased. When the load of information processing is concentrated locally, this becomes a bottleneck and reduces the throughput of the entire system. In order to avoid such load concentration, there is a load distribution technique for distributing the load (for example, Patent Document 1 and Patent Document 2).

  In recent years, with the aim of reducing CAPEX (CAPital EXpence) / OPEX (OPeration EXpence), the introduction of virtualization technology in communication networks has been promoted. As an example of the introduction of the virtualization technology, provision of a communication service with an N-Active redundant system configuration having high affinity with the virtualization technology and scaleability can be considered. Here, the N-Active redundant system configuration is a system configuration in which load distribution is performed by a plurality of devices or functional blocks that perform equivalent functions.

  Under such circumstances, it is possible to realize a highly reliable and scalable communication service by developing an application program using the distributed control platform middleware as shown in FIG.

  FIG. 1 shows an example of an architecture of a distributed control infrastructure. The architecture of the distributed control system shown in FIG. 1 employs a BDPS configuration (Blanacer, Dispatcher, Processor, Storage).

  In FIG. 1, 'C' represents a client terminal (hereinafter referred to as “client”). 'B' represents a load balancer that distributes data and processing requests sent from the client terminal C to a plurality of dispatchers (D) to reduce the load per unit. 'D' represents a dispatcher that allocates and allocates necessary resources for efficient processing when there are a plurality of data and processes waiting to be processed. 'P' represents an arithmetic processing unit (processor) that performs processing on an input signal. In the following, hardware or software that can be a processing subject including an application program (hereinafter referred to as “application”) executed by the processor may be indicated by “P”.

  When there is a request from the client terminal C regarding the signal distribution function of the distributed control system, the distributed control system performs the following distribution process. That is, B performs primary signal distribution (simple signal distribution). D performs distribution of secondary signals (guarantee of session uniqueness). In FIG. 1, the distributed control system performs signal distribution in a two-stage configuration.

  As a result, the bottleneck in B can be avoided, the scaleability of D can be guaranteed, and the signal distribution function unit itself can be prevented from becoming a bottleneck.

JP 2004-350188 A JP 2006-245841 A

  However, when a TCP connection path is newly established for a TCP communication whose direction is different from the TCP connection path in the direction from C to P (between CB, BD, and DP), it may be a resource depending on circumstances. May be consumed wastefully, leading to performance degradation of the distributed control system.

  For example, in such a distributed control system, there is a case where it is desired to transmit a TCP (Transmission Control Protocol) request from the distributed control system side toward the client. This will be described with reference to FIG.

  FIG. 2 is a diagram for explaining TCP connection path establishment in the distributed control infrastructure and TCP request transmission from the distributed control system side toward the client. When the communication service is provided by the TCP transport, when a TCP connection establishment is requested from the client (C) side, a TCP connection path is established in the direction from C to P. At that time, TCP connections are established between CB, BD, and DP, respectively.

  The application (P) on the distributed control infrastructure has an N-Active redundancy configuration. As shown in FIG. 2, the distributed control system receives a Push type TCP request (from P2 (for example, P2) different from P (for example, P1) that establishes a TCP connection with a specific client (C). There is a case where it is desired to transmit a TCP request from the distributed control system side toward the client (C).

  However, when a new connection path different from the TCP connection path in the direction from C to P (C-B, B-D, D-P) is newly established, resources are further consumed and the performance of the distributed control system is increased. Cause a decline.

  As one aspect of the present invention, when a connection is established between a processing request source and any one of a plurality of processing devices, effective utilization of resources when communicating from another processing device to the request source is achieved. Provide technology.

  A communication method in a load distribution system including a first distribution unit, a plurality of second distribution units, and a plurality of processing devices according to one aspect of the present invention performs the following processing. Here, the first distribution unit distributes each of the received plurality of processing requests. Each of the plurality of second distribution units further distributes each of the plurality of processing requests distributed by the first distribution unit. The plurality of processing devices respond to each of the plurality of processing requests distributed by any one of the plurality of second distribution units. Of the plurality of processing devices, the first processing device determines whether the first processing device and the processing request are in response to the first processing request distributed by any one of the plurality of second distribution units. A first connection is established with the request source. The second processing device among the plurality of processing devices communicates with the request source using a part of the established first connection in response to an instruction from a source other than the request source.

  According to an aspect of the present invention, when a connection is established between a processing request source and any one of a plurality of processing devices, effective use of resources when communicating from another processing device to the request source Can be achieved.

An example of an architecture of a distributed control infrastructure is shown. It is a figure for demonstrating TCP connection path establishment in a distributed control infrastructure, and TCP request transmission to the client direction from the distributed control system side. An example of the load distribution system in this embodiment is shown. It is a figure for demonstrating TCP request path | pass establishment in the distributed control infrastructure in this embodiment, and TCP request transmission to the client direction from the distributed control system side. It is a figure for demonstrating the TCP connection path establishment in a distributed control infrastructure in this embodiment. It is a figure for demonstrating the TCP connection management table in this embodiment. The distributed infrastructure system in the Example of this embodiment is shown. An example of each hardware constitutions of a processor (P), a dispatcher (D), and a load balancer (B) in the example of this embodiment is shown. It is an example of a system when a distributed control infrastructure is applied to a policy control server (PCRF) in an example of the present embodiment. A use case of a distributed control system based on a system configuration in an example of the present embodiment will be described. The format of the service signal used between the client (PCEF) -server (PCRF) in the Example of this embodiment is shown. An example of the TCP connection path establishment sequence in the Example of this embodiment is shown. FIG. 13 is a TCP connection management table used in the description of the sequence of FIG. 8 shows a Push type TCP request transmission sequence (part 1) in an example of the present embodiment. 8 shows a Push-type TCP request transmission sequence (part 2) in an example of the present embodiment. 15 is a TCP connection management table used in the description of the sequences of FIGS. 14A and 14B. 8 shows a Push type TCP response reception sequence in an example of the present embodiment. FIG. 17 is a TCP connection management table used in the description of the sequence of FIG.

  In the present embodiment, when a Push type TCP request is transmitted from an application (P) to a specific client (C) in an N-Active redundant configuration distributed control system constructed with a distributed control infrastructure, the following processing is performed. Do. That is, a Push type TCP request is transmitted to the client (C) via the TCP connection established with the client (C).

  FIG. 3 shows an example of a load distribution system in the present embodiment. The load distribution system 1 includes a first distribution unit 2, a plurality of second distribution units 3, and a plurality of processing devices 4.

The first distribution unit 2 distributes each of the received plurality of processing requests.
Each of the plurality of second distribution units 3 further distributes each of the plurality of processing requests distributed by the first distribution unit 2.

  The plurality of processing devices 4 respond to each of the plurality of processing requests distributed by any one of the plurality of second distribution units 3.

  The first processing device 4-1 among the plurality of processing devices 4 responds to the first processing request distributed by one of the second distribution units 3-1 of the plurality of second distribution units 3. A first connection is established between the first processing device 4-1 and the request source 5 of the processing request.

  The second processing device 4-2 among the plurality of processing devices 4 communicates with the request source 5 using a part of the established first connection in response to an instruction from other than the request source.

  With this configuration, when a connection is established between a processing request source and one of a plurality of processing devices, effective use of resources when communicating from another processing device to the request source is achieved. Can be planned.

  The second processing device 4-2 assigns the second distribution unit 3-1 and the first distribution unit 2 that distribute the first processing request to the first processing device 4-1, in response to an instruction from other than the request source. The communication with the request source 5 is performed using a part of the established first connection.

  With this configuration, it is possible to communicate with the request source by effectively utilizing the established first connection.

  The second processing device 4-2 acquires request source information that identifies the request source 5 from an instruction from a source other than the request source. Based on the request source information, the second processing device 4-2 inquires the plurality of second distribution units whether they have the first connection established with the request source 5. The second processing device 4-2 sends a part of the established first connection to the request source 5 via the second distribution unit 3-1, which has the established first connection, according to the result of the inquiry. To communicate with the request source 5.

  With this configuration, the request source 5 and the second distribution unit 3-1 having the established first connection are identified, and the request source 5 and the request source 5 are identified using a part of the established first connection. Communication can be performed.

  Each of the plurality of second distribution units 3 determines whether or not the requester 5 has an established first connection in response to an inquiry from the second processing device 4-2. When each of the plurality of second distribution units 3 has an established first connection, it is used among the plurality of second connections between the second distribution unit 3-1 itself and each of the plurality of processing devices. A second connection that is not yet selected is selected. Each of the plurality of second distribution units 3 associates the selected second connection with a part of the established first connection.

  With this configuration, the second processing device 4-2 can communicate with the request source 5 using a part of the established first connection.

  The second distribution unit 3-1 having the established first connection with the request source acquires request information addressed to the request source from the second processing device via the selected second connection. The second distribution unit 3-1 extracts service identification information that uniquely identifies the communication service from the request information. The second distribution unit 3-1 holds connection management information in which the extracted service identification information is associated with the second connection identification information for identifying the selected second connection, and transmits the request information to the request source 5. .

  With this configuration, the second distribution unit 3 can identify to which processing device the response to the processing request to the request source 5 is transferred.

  Each of the plurality of second distribution units 3 acquires the communication information distributed by the first distribution unit 2. The second distribution unit 3 extracts service identification information from the communication information. The second distribution unit 3 acquires second connection identification information having service identification information that matches the extracted service identification information from the connection management information. The second distribution unit 3 transmits the communication information to the second processing device 4 via the second connection corresponding to the acquired second connection identification information.

  By configuring in this way, the response information to the request information for the request source 5 can be transferred to the processing apparatus that has transmitted the request.

  Each of the plurality of second distribution units 3 holds connection management information. The connection management information is information in which the second connection identification information, the usage state information, the third connection identification information, the service identification information, and the request source identification information are associated with each other. The second connection identification information identifies a second connection between the second distribution unit itself and each of the plurality of processing devices. The usage state information indicates whether the second connection is being used. The third connection identification information is a third connection using the second connection, and identifies the third connection between the second distribution unit itself and the first distribution unit. The service identification information uniquely identifies the communication service between the second processing device 4-2 and the request source 5. The request source specifying information holds that the request source is specified.

  FIG. 4 is a diagram for explaining TCP connection path establishment in the distributed control infrastructure and TCP request transmission from the distributed control system side toward the client in the present embodiment.

  The configuration of the distributed control system 11 is the same as the configuration described in FIG. When the communication service is provided by the TCP transport, when a TCP connection establishment request is made from the client (C) 15 side, a TCP connection path is established in the direction from the client (C) 15 to the processor (for example, P1) 12. At that time, TCP connections are established between the client (C) 15 and the load balancer (B) 14, between the load balancer (B) 14 and the dispatcher (D) 13, and between the dispatcher (D) 13 and the processor (P1) 12. .

  At this time, when the application (P2) 12 transmits a push-type TCP request to the client (C) 15, the application (P2) 12 transmits a push-type TCP request using a TCP connection established between C and P1. .

  Hereinafter, the processor is referred to as 'P'. The dispatcher is called 'D'. The load balancer is called “B”. The client is called 'C'.

  In the present embodiment, as an example, the distributed control system 1 distributes signals in a two-stage configuration of B and a plurality of Ds, but is not limited to this. For example, a plurality of Ds are arranged in multiple layers. It may be configured.

  For example, in the case of signal distribution in a three-stage configuration, B in the first layer distributes the input signal to one of a plurality of Ds in the second layer. One of the plurality of Ds in the second layer further distributes the signal distributed by B to one of the plurality of Ds in the third layer. One of the plurality of Ds in the third layer further distributes the signal distributed by D in the second layer to any of the plurality of Ps.

  FIG. 5 is a diagram for explaining the TCP connection path establishment in the distributed control infrastructure in the present embodiment. D has a TCP connection management table 21.

  FIG. 6 is a diagram for explaining a TCP connection management table in the present embodiment. FIG. 6A shows an example of the TCP connection management table 21 when a TCP connection establishment request is made from the client side. FIG. 6B shows an example of the TCP connection management table 21 when a TCP request is made from the distributed control system side toward the client in the TCP connection established state of FIG. The TCP connection management table 21 is managed by D13.

  The TCP connection management table includes fields of “P-D connection POOL”, “Usage status (B-D connection ID)”, and “Client information (IP address & port number)”.

  “P-D connection POOL” is a field for a connection between P-D. The “P-D connection POOL” field includes fields of “P address”, “P port”, “P-D connection ID”, and “Key character string”.

  The “P address” field stores the address of the processor (P). The “P port” field stores port information of the processor (P) used for connection between PD. “P-D connection ID” stores identification information for identifying a P-D connection.

  “Key Character String” confirms a unique character string (L7 information) for the service. When D receives a Push-type TCP request signal from P, D extracts L7 information from the received signal and stores it in the TCP connection management table 21. The unique character string (L7 information) is used for D to identify which P the response signal for the Push-type TCP request is assigned to. L7 means the seventh layer of the OSI reference model.

  Information registered in each field of “P-D connection POOL” is registered in the D and P server restart initial setting process.

  The “usage state” field stores state information indicating whether the connection between P-D specified by the connection ID between P-D is in use or free. In the “use state” field, when the connection between P-D is in use, the ID of the connection between B-D using the connection between P-D is further stored. In the “use state” field, information indicating a “busy” state is set when a TCP connection establishment request is issued from C, and information indicating a “free” state is set when a TCP connection disconnection request is triggered.

  “Client information” stores the IP (Internet Protocol) address and port number of the client. When a TCP request is received from the client after the TCP connection is established from the client 15, D extracts client information as L7 information from the received signal and stores it in the TCP connection management table 21. The client information is used when specifying D having a TCP connection that P has already established.

  Hereinafter, FIG. 6 will be described with reference to FIG. The connection between P and D is established in the direction from D to P when D and P are activated and restarted. For the connection between PDs, a plurality of TCP connections are established by the distributed control infrastructure. In D, when a TCP connection establishment request is issued from C, an empty D-P connection is selected from “D-P connection POOL”, and information indicating that it is in use is set in the use state field. Thereby, as shown in FIG. 6A, the association between the BD connection and the DP connection is managed.

  Furthermore, P2 (12) that wants to send a Push-type TCP request to a specific client (C) searches the TCP connection management table 21 (FIG. 6A) using the client information of the specific client. That is, P2 (12) specifies a TCP connection with the already established client from the TCP connection management table 21 (FIG. 6A). At this time, D13 selects a free connection from the P-D connection POOL with P2 (12), and updates the selected connection to the “in use” state in the TCP connection management table 21 ( FIG. 6 (B)).

  Thereafter, P2 (12) uses the selected connection to send a Push-type TCP request to a specific client. At this time, D33 extracts unique information such as a session ID from the Push-type TCP request, and registers it in the TCP connection management table 21 as a Key character string (FIG. 6B). Thereby, the response corresponding to the Push type TCP request can be specified.

  As described above, the TCP connection management table 21 managed by the D 13 is provided with fields for managing client information (IP address & port number) and a service unique key character string.

  Thereby, P can grasp | ascertain which D has already established TCP connection with a specific client (C) by using client information. As a result, P can specify D having a TCP connection.

  Further, by using the Key character string, D can identify to which P the response signal for the Push-type TCP request for the specific client (C) is to be distributed. As a result, the Push type TCP request transaction in D can be managed.

  According to this embodiment, when a connection between C and P1 is established via B and D, P2 communicates with C using the established connection, so that resources can be effectively used.

Examples of the present embodiment will be described below.
FIG. 7 shows a distributed infrastructure system in an example of the present embodiment. In FIG. 7, the distributed infrastructure system 31 and the client (C) 35 are connected via a network. The client (C) 35 is a client device that uses a service built on the distributed control infrastructure. The network may be a communication network such as the Internet, a LAN, a WAN, a dedicated line, a wired line, and a wireless line.

  The distributed infrastructure system 31 includes a server (P (P1, P2)) 32, a dispatcher (D) 33, and a load balancer (B) 34. A plurality of connections can be established between PDs (P-D connection POOL).

  The load balancer (B) 34 is a load distribution device, and distributes primary signals for external access signals. In the case of TCP transport, B34 performs signal distribution in connection units. When a TCP connection is established in response to a request from C35, B34 manages association information with the C-B TCP connection and the B-D TCP connection, and performs signal distribution in connection units.

  The dispatcher (D) 33 performs secondary signal distribution on the signal distributed at B34. D33 extracts the service unique key character string information from the received L7 signal, and determines the signal distribution destination server, that is, P from the key character string information. When the signal transport is TCP, D33 manages the relationship information between the BD connection and the PD connection in the TCP connection management table 41 as described above.

  Also, D33 guarantees session uniqueness regarding the Push-type TCP request signal by managing service-specific key character string information in the key character string field of the TCP connection management table 41.

  The server (P (P1, P2)) 32 is a server on which application logic is mounted. When transmitting a Push-type TCP request signal, P (P1, P2) 32 specifies D having a TCP connection already established with C35 and transmits the Push-type TCP request signal.

  Distributed control infrastructure middleware 42 is provided in D33 and P32. The distributed control infrastructure middleware 42 provides a function for guaranteeing application operation as a distributed control system having an N-Active redundancy configuration.

  FIG. 8 shows an example of the hardware configuration of each of the processor (P), the dispatcher (D), and the load balancer (B) in the example of the present embodiment. Each of P32, D33, and B34 includes an interface device 51, a CPU (Central Processing Unit) 52, a memory 53, a disk 54, and a bus 55.

  The bus 55 is a bus for transferring information among the interface device 51, the CPU 52, the memory 53, and the disk 54.

  The interface device 51 is an interface for performing input / output with respect to each of P32, D33, and B34. The memory 53 is a volatile storage device that temporarily holds information.

  The disk 54 is a storage device that stores a large amount of information. As the disk 54, various types of storage devices such as a hard disk and a flash memory card can be used.

  The CPU 52 is a central processing unit that performs overall control for each of P32, D33, and B34. The CPU 52 reads the program according to the present embodiment from the disk 54 and executes the program.

  The program that realizes the processing described in the above embodiment may be stored in, for example, the disk 54 via the communication network and the interface device 51 from the program provider side. Moreover, the program which implement | achieves the process demonstrated by the said embodiment may be stored in the portable storage medium marketed and distribute | circulated. In this case, the portable storage medium may be set in a reading device (not shown) and the program read by the CPU 52 and executed. As a portable storage medium, various types of storage media such as a CD-ROM, a flexible disk, an optical disk, a magneto-optical disk, an IC card, a USB (Universal Serial Bus) memory device, and a semiconductor memory card can be used.

  FIG. 9 shows an example of a system when a distributed control infrastructure is applied to the policy control server (PCRF) in the example of the present embodiment.

  The policy execution server (PCEF) 61 is a server that monitors and restricts service traffic (such as an upper limit on the number of packets per month) according to a policy for each subscriber. The policy execution server 61 is a client (counter device) of the policy control server (PCRF) 62.

  The policy control server (PCRF) 62 is a server that distributes policy information for each subscriber to the policy execution server 61. The policy information distribution trigger is an update timing of policy information by a maintenance person or an opportunity of receiving a distribution request from the policy execution server 61. In the present embodiment, the PCRF 62 is a server constructed with a distributed control infrastructure, that is, the distributed processing system 31 in FIG.

  The policy information database (SPR) 63 is a server that holds policy information. When the maintenance person updates the policy information, the policy update information is notified to the PCRF 62.

  Here, the communication protocol used between the client (PCEF) 61 and the server (PCRF) 62 and between the policy information database (SPR) 63 and the server (PCRF) 62 is DIAMETER (RFC3588, RFC4006 recommendation). However, in the 3GPP recommendation, PCEF-PCRF is specified as a Gx interface, and PCRF-SPR is specified as a separate interface and an Sp interface (3GPP TS 23.203, TS 29.213 recommendation).

  In the present embodiment, an interface used between the client (PCEF) 61 and the server (PCRF) 62 is represented by an interface A. An interface used between the policy information database (SPR) 63 and the server (PCRF) 62 is represented by an interface B.

  FIG. 10 shows a use case of the distributed control system based on the system configuration in the example of the present embodiment.

  Since the interface between the client (PCEF) 61 and the server (PCRF) 62 and the interface between the policy information database (SPR) 63 and the server (PCRF) 62 are different, when the signal is distributed to the same P32 in the server (PCRF) 62 Is not limited.

  Assume that P1 receives a service signal related to interface A and P2 receives a service signal related to interface B. In this case, there is a case where P2 wishes to transmit a Push type TCP request signal to the client (PCEF) 61 when the latter service signal is received.

  For example, as described above, it is assumed that the PCRF 62 notifies the PCEF 61 of the updated policy information when the maintenance information is updated for a subscriber in SPR63. In this case, the PCRF 62 transmits a Push-type TCP request signal to the PCEF 61 via the TCP connection established between the PCEF 61 and the PCRF 62.

  FIG. 11 shows a format of a service signal used between the client (PCEF) and the server (PCRF) in the example of the present embodiment.

  Credit-Control-Request (CCR) (FIG. 11A), Credit-Control-Answer (CCA) (FIG. 11B), Re-Auth-Request (RAR) (FIG. 11C), Re- Each of Auth-Answer (RAA) (FIG. 11D) includes items of “session ID” and “origin-host”. “Session ID” is information for identifying a session between the client (PCEF) 61 and the server (PCRF) 62. “Origin-host” is information about client information.

  At the start of connection from the PCEF 61 to the PCRF 62, a TCP connection is established in the CCR or CCA service signal processing.

  In the signal distribution logic in D, for example, D extracts the CCR “session ID” as a service-unique key character string and specifies P (application) of the signal distribution destination. In the present embodiment, information managed in the client information field of the TCP connection management table 41 by D is, for example, “Origin-Host” of CCR.

  When the policy information update is transmitted from the PCRF 62 to the PCEF 61, an RAR or RAA service signal is used.

  The Push-type TCP request signal mentioned so far corresponds to, for example, this RAR signal, and is a request signal transmitted from the server (P) to the client (C). In this embodiment, the information that D manages in the key character string field of the TCP connection management table 41 is, for example, key character string information of the “session ID” of RAR (which may be a hash value in terms of mounting).

  FIG. 12 shows an example of a TCP connection path establishment sequence in the example of the present embodiment. FIG. 13 is a TCP connection management table used in the description of the sequence of FIG.

  D33 receives a TCP request indicating CCR from C35 through B34 (S1, S2). Then, D33 extracts the session ID from the CCR as a key character string (S3). Further, D33 extracts “Origin-Host” as client information from the CCR.

  D33 determines the signal transfer destination P32 from the Key character string (S4). For example, it is assumed that D33 has correspondence information in which a hash value and P are associated with each other. D33 generates a hash value from the Key character string using the hash function, and selects P corresponding to the generated hash value using the correspondence information.

  D33 refers to the TCP connection management table 41 and selects an unused TCP connection from the signal transfer destination P and the established inter-P-D connection pool when the server is restarted (S5).

  D33 associates the selected unused TCP connection with the B (C) -D TCP connection used in CCR, and updates the TCP connection management table 41 (S6). Specifically, the D33 registers the B (C) -D TCP connection ID and client information on the TCP connection management table 41 as shown in FIG.

  When D33 receives a TCP request from B (C), D-1 sets “−1” in the Key character string field of the TCP connection management table 41 related to the selected P-D connection as shown in FIG. To do.

  D33 transmits a CCR signal to P32 using the selected P-D connection (S7, S8). When P32 receives the CCR signal from D33, P32 returns a CCA signal (S9).

  D33 specifies the P-D connection used for receiving the CCA signal in the TCP connection management table 41 as the B-D connection managed in association (S10).

D33 transmits a CCA signal to C35 via B34 (S12, S13).
14A and 14B show a Push-type TCP request transmission sequence in the example of the present embodiment. FIG. 15 is a TCP connection management table used in the description of the sequences of FIGS. 14A and 14B.

  In order for P2 (32) to send a Push-type TCP request to the client (C) 35 when the policy information is updated from the SPR 63, first, which D has an already established TCP connection with C. It is necessary to identify whether it is doing. Therefore, with client information (IP address & port number) as input, P2 (32) inquires of all D whether each D already has a TCP connection established with the client.

  Specifically, P2 (32) receives from the SPR 63 a notification that the policy information has been updated (policy information update notification) (S21). P2 (32) extracts client information from the received policy information update notification (S22).

  P2 (32) multicasts a D search signal to all D33s, and searches for a D for which a TCP connection has been established from the TCP connection management table 41 of each D33 based on the extracted client information (S23). , S24).

  Each D33 receives a D search signal from P. Then, D33 searches the TCP connection management table 41 (FIG. 15A) using the client information (IP address & port number) included in the D search signal, and D itself has already been established with the client. It is determined whether there is a TCP connection (S25).

  In the TCP connection management table 41 (FIG. 15A), D33 selects an empty connection from the P-D connection pool with P2 (32), which is the transmission source P of the D search signal. As shown in FIG. 15B, D33 sets the “in use” state in the use state field of the selected free connection (S26). D33 returns the search result to P2 (32) as a response signal to the D search signal (S27, S28).

  When P2 (32) receives a response signal to the D search signal, P2 (32) specifies D having a TCP connection with the client, and transmits a Push-type TCP request signal via the specified D.

  Specifically, P2 (32) specifies the connection between P and D from the IP address and port information of D that is notified by the response signal to the D search signal (S29). P2 (32) transmits an RAR signal as a Push-type TCP request signal indicating the policy information update notification to D33 (S30).

  D33 receives the Push-type TCP request signal (L7 signal) from P2 (32). Then, D33 extracts key character string information unique to the service from the received signal information. As shown in FIG. 15C, D33 sets the extracted key character string information in the key character string field relating to the P-D TCP connection in the TCP connection management table 41 (S31, S33).

  Further, D33 searches the TCP connection management table using the TCP connection ID of the P-D TCP connection used for receiving the signal information, and specifies the associated B (C) -D TCP connection ID. (S32).

  The D33 searches the TCP connection management table 41 using the received P-D connection and identifies the associated BD connection (S34).

  D33 transmits a Push type TCP request signal to a desired client (C) using the B (C) -D TCP connection specified in S34 (S35, S36, S37).

  FIG. 16 shows a Push-type TCP response reception sequence in the example of the present embodiment. FIG. 17 is a TCP connection management table used in the description of the sequence of FIG.

  D33 receives the CCR as a Push type TCP response signal from the client (C) 35 (S41, S42).

  D33 searches the TCP connection management table 41 using the TCP connection ID of the B (C) -D TCP connection used for receiving the CCR, and specifies one or more associated P-D TCP connection IDs. To do. Here, if a Push type TCP request signal is simultaneously transmitted from a plurality of Ps to a specific client (C), a plurality of P-D TCP connection IDs may be extracted. .

  D33 extracts the session ID as a Key character string from the received CCR (S43). D33 specifies a TCP connection ID between P-Ds having the same service unique key character string information ("XXXX") among the TCP connections between P-Ds extracted in S44 from the TCP connection management table 41 ( S44).

  D33 transmits a Push type TCP response signal (RAA) to P2 (32) using the specified P-D TCP connection (S45, S46). As a result, the response signal reaches the source P2 (32) of the Push type TCP request signal.

  D33 deletes the Key character string information (“XXXX”) used in S44 from the Key character string field of the TCP connection management table 41 (S47).

  If the Key character string field corresponding to the P-D connection used for the transmission of the RAA signal is empty, D33 also clears the usage state (BD connection ID) and the client information field (S48). ).

  According to the present embodiment, when a Push-type TCP request signal is transmitted from an application on the distributed control infrastructure, it is possible to transmit it using an established TCP connection. Therefore, when the application (P) having an N-Active redundancy configuration transmits a Push-type TCP request signal, the processing cost (processing load and software resource capture) for establishing a new TCP connection can be reduced in P and B. is there.

  The present invention is not limited to the above-described embodiment, and various configurations or embodiments can be taken without departing from the gist of the present invention.

Regarding the above embodiment, the following additional notes are disclosed.
(Appendix 1)
A first distribution unit that distributes each of the plurality of received processing requests;
A plurality of second distribution units that further distribute each of the plurality of processing requests distributed by the first distribution unit;
A communication method in a load distribution system including a plurality of processing devices that respond to each of the plurality of processing requests distributed by any of the plurality of second distribution units,
The first processing device of the plurality of processing devices, in response to the first processing request distributed by any one of the plurality of second distribution units, the first processing device and Establishing a first connection with the requester of the processing request;
The second processing device among the plurality of processing devices communicates with the request source using a part of the established first connection in response to an instruction from a source other than the request source. Communication method.
(Appendix 2)
The second processing device is established via the second distribution unit and the first distribution unit that distribute the first processing request to the first processing device in response to an instruction from a source other than the request source. The communication method according to claim 1, wherein a part of the first connection that has already been used is used to communicate with the request source.
(Appendix 3)
The second processing device acquires request source information for specifying the request source from an instruction from a source other than the request source, and requests the plurality of second distribution units based on the request source information. The first connection already established and the first connection established are inquired, and the established first connection is made via the second distribution unit having the first connection established with the request source in accordance with the result of the inquiry. The communication method according to appendix 1, wherein a part of the connection is used to communicate with the request source.
(Appendix 4)
Each of the plurality of second allocating units determines whether the requester has the first connection established in response to the inquiry from the second processing device, and the established first connection In the case of having a connection, a second connection that is not used among a plurality of second connections between the second distribution unit itself and each of the plurality of processing devices is selected, and the selected second connection and the The communication method according to appendix 3, wherein a part of the established first connection is associated.
(Appendix 5)
The second distribution unit having the first connection established with the request source acquires request information addressed to the request source from the second processing device via the selected second connection, and the request Service identification information for uniquely identifying a communication service is extracted from the information, and connection management information relating the extracted service identification information to second connection identification information for identifying the selected second connection is retained, The communication method according to appendix 4, wherein the request information is transmitted to the request source.
(Appendix 6)
Each of the plurality of second distribution units acquires the communication information distributed by the first distribution unit, extracts the service identification information from the communication information, and extracts the service from the connection management information The second connection identification information having the service identification information that matches the identification information is acquired, and the communication information is transmitted to the second processing device via the second connection corresponding to the acquired second connection identification information. The communication method according to appendix 5, characterized in that:
(Appendix 7)
Each of the plurality of second distribution units uses second connection identification information for identifying a second connection between the second distribution unit itself and each of the plurality of processing devices, and the second connection uses the second connection identification information. The use state information indicating whether the second connection is made, and the third connection using the second connection and identifying the third connection between the second distribution unit itself and the first distribution unit A connection in which third connection identification information, service identification information for uniquely identifying a communication service between the second processing device and the request source, and request source identification information for identifying the request source are associated with each other. The communication method according to attachment 1, wherein management information is held.
(Appendix 8)
A first distribution unit that distributes each of the plurality of received processing requests;
A plurality of second distribution units that further distribute each of the plurality of processing requests distributed by the first distribution unit;
A plurality of processing devices that respond to each of the plurality of processing requests distributed by any of the plurality of second distribution units;
With
The first processing device of the plurality of processing devices, in response to the first processing request distributed by any one of the plurality of second distribution units, the first processing device and Establishing a first connection with the requester of the processing request;
The second processing device among the plurality of processing devices communicates with the request source using a part of the established first connection in response to an instruction from a source other than the request source. Load balancing system.
(Appendix 9)
The second processing device is established via the second distribution unit and the first distribution unit that distribute the first processing request to the first processing device in response to an instruction from a source other than the request source. The load balancing system according to appendix 8, wherein a part of the first connection that has already been used is used for communication with the request source.
(Appendix 10)
The second processing device acquires request source information for specifying the request source from an instruction from a source other than the request source, and requests the plurality of second distribution units based on the request source information. The first connection already established and the first connection established are inquired, and the established first connection is made via the second distribution unit having the first connection established with the request source in accordance with the result of the inquiry. The load balancing system according to appendix 8, wherein a part of the connection is used to communicate with the request source.
(Appendix 11)
Each of the plurality of second allocating units determines whether the requester has the first connection established in response to the inquiry from the second processing device, and the established first connection In the case of having a connection, a second connection that is not used among a plurality of second connections between the second distribution unit itself and each of the plurality of processing devices is selected, and the selected second connection and the The load balancing system according to appendix 10, wherein a part of the established first connection is associated.
(Appendix 12)
The second distribution unit having the first connection established with the request source acquires request information addressed to the request source from the second processing device via the selected second connection, and the request Service identification information for uniquely identifying a communication service is extracted from the information, and connection management information relating the extracted service identification information to second connection identification information for identifying the selected second connection is retained, The load distribution system according to appendix 11, wherein the request information is transmitted to the request source.
(Appendix 13)
Each of the plurality of second distribution units acquires the communication information distributed by the first distribution unit, extracts the service identification information from the communication information, and extracts the service from the connection management information The second connection identification information having the service identification information that matches the identification information is acquired, and the communication information is transmitted to the second processing device via the second connection corresponding to the acquired second connection identification information. Item 13. The load distribution system according to appendix 12, wherein
(Appendix 14)
Each of the plurality of second distribution units uses second connection identification information for identifying a second connection between the second distribution unit itself and each of the plurality of processing devices, and the second connection uses the second connection identification information. The use state information indicating whether the second connection is made, and the third connection using the second connection and identifying the third connection between the second distribution unit itself and the first distribution unit A connection in which third connection identification information, service identification information for uniquely identifying a communication service between the second processing device and the request source, and request source identification information for identifying the request source are associated with each other. The load balancing system according to appendix 8, wherein management information is retained.

DESCRIPTION OF SYMBOLS 1 Load distribution system 2 1st distribution part 3 (3-1) 2nd distribution part 4 Processing apparatus 4-1 1st processing apparatus 4-2 2nd processing apparatus 11 Distributed control system 12 Processor (P, P1, P2) )
13 Dispatcher (D)
14 Load balancer (B)
15 Client (C)
21 TCP connection management table 31 Distributed infrastructure system 32 Server (P (P1, P2))
33 Dispatcher (D)
34 Load Balancer (B)
35 Client (C)
41 TCP connection management table 51 Interface device 52 CPU
53 Memory 54 Disk 55 Bus 61 Policy execution server (PCEF)
62 Policy Control Server (PCRF)
63 Policy Information Database (SPR)

Claims (8)

A first distribution unit that distributes each of the plurality of received processing requests;
A plurality of second distribution units that further distribute each of the plurality of processing requests distributed by the first distribution unit;
A communication method in a load distribution system including a plurality of processing devices that respond to each of the plurality of processing requests distributed by any of the plurality of second distribution units,
The first processing device of the plurality of processing devices, in response to the first processing request distributed by any one of the plurality of second distribution units, the first processing device and Establishing a first connection with the requester of the processing request;
The second processing device among the plurality of processing devices communicates with the request source using a part of the established first connection in response to an instruction from a source other than the request source. Communication method. The second processing device is established via the second distribution unit and the first distribution unit that distribute the first processing request to the first processing device in response to an instruction from a source other than the request source. The communication method according to claim 1, wherein communication is performed with the request source by using a part of the first connection that has been completed. The second processing device acquires request source information for specifying the request source from an instruction from a source other than the request source, and requests the plurality of second distribution units based on the request source information. The first connection already established and the first connection established are inquired, and the established first connection is made via the second distribution unit having the first connection established with the request source in accordance with the result of the inquiry. The communication method according to claim 1, wherein a part of the connection is used to communicate with the request source. Each of the plurality of second allocating units determines whether the requester has the first connection established in response to the inquiry from the second processing device, and the established first connection In the case of having a connection, a second connection that is not used among a plurality of second connections between the second distribution unit itself and each of the plurality of processing devices is selected, and the selected second connection and the The communication method according to claim 3, wherein a part of the established first connection is associated. The second distribution unit having the first connection established with the request source acquires request information addressed to the request source from the second processing device via the selected second connection, and the request Service identification information for uniquely identifying a communication service is extracted from the information, and connection management information relating the extracted service identification information to second connection identification information for identifying the selected second connection is retained, The communication method according to claim 4, wherein the request information is transmitted to the request source. Each of the plurality of second distribution units acquires the communication information distributed by the first distribution unit, extracts the service identification information from the communication information, and extracts the service from the connection management information The second connection identification information having the service identification information that matches the identification information is acquired, and the communication information is transmitted to the second processing device via the second connection corresponding to the acquired second connection identification information. The communication method according to claim 5. Each of the plurality of second distribution units uses second connection identification information for identifying a second connection between the second distribution unit itself and each of the plurality of processing devices, and the second connection uses the second connection identification information. The use state information indicating whether the second connection is made, and the third connection using the second connection and identifying the third connection between the second distribution unit itself and the first distribution unit A connection in which third connection identification information, service identification information for uniquely identifying a communication service between the second processing device and the request source, and request source identification information for identifying the request source are associated with each other. Management information is held. The communication method according to claim 1 characterized by things. A first distribution unit that distributes each of the plurality of received processing requests;
A plurality of second distribution units that further distribute each of the plurality of processing requests distributed by the first distribution unit;
A plurality of processing devices that respond to each of the plurality of processing requests distributed by any of the plurality of second distribution units;
With
The first processing device of the plurality of processing devices, in response to the first processing request distributed by any one of the plurality of second distribution units, the first processing device and Establishing a first connection with the requester of the processing request;
The second processing device among the plurality of processing devices communicates with the request source using a part of the established first connection in response to an instruction from a source other than the request source. Load balancing system.

Images