JP5793126B2 - Cluster system - Google Patents

Description

  The present invention relates to a technique for specifying an apparatus related to a complaint such as a defect report from a user in a scale-out type cluster system.

In recent years, a scale-out type cluster system capable of flexibly changing a configuration in response to a failure or a demand fluctuation has been studied (for example, Non-Patent Document 1). This cluster system includes a load balancer, a distribution device, and a signal processing device.
The load balancer has a function of receiving a message from a client terminal used by a user, distributing the received message to one of a plurality of distribution devices, and transmitting the message. The distribution device has a function of distributing a received message to any one of a plurality of signal processing devices using a unique value included in the message as a key. The signal processing device is, for example, a server, and has a function of generating and returning response information corresponding to the received message.

  Non-Patent Document 1 discloses a technique that uses a consistent hashing method as a sorting method of a sorting device and facilitates a configuration change of the signal processing device. And it has been shown that by making it possible to easily change the number of signal processing units, it is possible to realize a scale-out type cluster system that can flexibly change the configuration in response to failures and demand fluctuations. .

Michio Irie, 4 others, "Cluster model of session control server that realizes scale-out and flexible configuration change", IEICE, IEICE, March 2011, B-6-11

  However, in Non-Patent Document 1, when a user has a complaint regarding a defect or the like, in order to confirm the state of the defect corresponding to the complaint, investigate the cause, and recover, the apparatus related to the complaint is quickly identified. The technology is not described. In particular, in a scale-out type cluster system that provides a communication service that requires high reliability and availability, it is important to quickly identify a device related to a claim.

  Therefore, an object of the present invention is to provide a technique for quickly identifying a device related to a complaint such as a defect in a scale-out type cluster system.

  The present invention is a cluster system including a set composed of a plurality of signal processing devices, and extracts a message from messages of the same session, and a user who transmits and receives the message included in the extracted message. A set of the signal processing device and a new signal determination unit that transmits a user-specific information indicating information to be identified and a mapping value calculated by applying a unique value included in the message to a predetermined function as a log to the history detection unit; When changed, a signal processing device ID indicating information for identifying a signal processing device constituting the changed set, and a mapping value associated with the signal processing device ID as configuration change information to the history detection means A distribution destination determining means for transmitting, receiving the log and storing it in log information of a storage means; Is received and stored in the distribution destination history information of the storage means, the user specifying information input from a terminal is acquired, and the log that matches the acquired user specifying information is obtained by referring to the log information. The first entry shown is extracted, the mapping value included in the first entry is referenced using the mapping value included in the first entry, and the mapping value included in the first entry is stored in the distribution destination history information And a history search unit that identifies the signal processing device ID when the value matches the mapping value.

  According to such a configuration, the cluster system includes the new signal determination unit that acquires the log of the message and the distribution destination determination unit that acquires the history of the set of the signal processing device, and the history search unit acquires Using the user identification information, it is possible to quickly identify the signal processing apparatus ID associated with the user identification information by referring to the log information and the distribution destination history information. That is, the cluster system can quickly narrow down the signal processing devices related to a complaint when the user has a complaint regarding a defect or the like.

  Further, the new signal determination unit extracts one message from messages in the same session and stores the log in the log information. Therefore, since the log information stores the minimum number of logs for each session, it is possible to relatively reduce the load related to log transmission / storage as compared with the case where logs of all messages are stored. it can. In addition, since the number of logs stored in the log information is relatively small, the reference process by the history search unit can be quickly terminated. That is, the cluster system can quickly narrow down the signal processing devices related to a complaint when the user has a complaint regarding a defect or the like.

  In addition, since the cluster system has distribution destination history information, even if the signal processing device set is changed during the same session, the signal processing device of the message distribution destination is tracked. Can do. In other words, regardless of whether the set of signal processing devices is changed, the cluster system can quickly narrow down the signal processing devices related to the complaint when the user has a complaint regarding a defect or the like.

  In the present invention, the new signal determination unit further transmits the extracted message including the time when the message was transmitted, and the distribution destination determination unit further indicates the time when the set was changed as the configuration change information. When the history search means includes a plurality of extracted first entries, the set is configured based on the time when the set is changed by receiving the configuration change information. A distribution time range indicating a range of times is generated and stored in the distribution destination history information in association with a signal processing device ID indicating information for identifying the signal processing devices constituting the set. Obtaining the second entry by extracting the first entry at a time closest to the reporting time before the acquired reporting time from the plurality of the first entries. Referring to the distribution destination history information using a set of time and mapping value included in the second entry, the time is included in the distribution time range, and the mapping value is stored in the distribution destination history information The signal processing apparatus ID is specified when the mapping value matches the mapping value.

  According to such a configuration, the cluster system can specify the signal processing device that has executed the message processing related to the user specifying information at the time closest to the time before the newly acquired time. There is a high possibility that a cause related to the complaint from the user has occurred at the time closest to the time before the acquired time. Therefore, the cluster system can quickly identify a signal processing device that is highly likely to be related to a complaint from the user.

  The present invention is a cluster system including a set composed of a plurality of signal processing devices, and one message is extracted from messages of the same session, and the time when the extracted message is transmitted is included in the message. New signal determination means for transmitting user identification information indicating information for identifying a user who transmits and receives the message and a mapping value calculated by applying a unique value included in the message to a predetermined function as a log to the history detection means; When the set of signal processing devices is changed, the time at which the set is changed, the signal processing device ID indicating information for identifying the signal processing devices constituting the changed set, and the signal processing device ID A distribution destination determination means for transmitting the attached mapping value to the history detection means as configuration change information; The log is received and stored in the log information of the storage means. Based on the time when the set was changed by receiving the configuration change information, a distribution time range indicating the range of time when the set was configured is generated. Then, the distribution destination history information of the storage means in association with the signal processing device ID indicating information for identifying the distribution processing time range and the signal processing device constituting the set, and the mapping value associated with the signal processing device ID And the user specific information acquired from the terminal is acquired, the log information is referred to, and the acquired user specific information matches within a predetermined period before the acquired report time. The first entry indicating the log to be extracted is extracted, and the distribution destination history information is referred to using a set of time and mapping value included in the first entry. And a history search means for identifying the signal processing device ID when the time is included in the distribution time range and the mapping value matches the mapping value stored in the distribution destination history information, It is characterized by providing.

  According to such a configuration, the cluster system refers to the log information and the distribution destination history information using the acquired reporting time and the user specifying information, and stores the user specifying information within a predetermined period before the acquired reporting time. It is possible to quickly specify the associated signal processing device ID. That is, the cluster system can quickly narrow down the signal processing devices related to a complaint when the user has a complaint regarding a defect or the like.

  In the present invention, when the history search means includes a plurality of the extracted first entries, the history search means has a time closest to the reporting time before the acquired reporting time from among the plurality of first entries. The first entry is extracted as a second entry, the distribution destination history information is referred to using the combination of the time and the mapping value included in the second entry, the time is included in the distribution time range, and The signal processing device ID is specified when the mapping value matches the mapping value stored in the distribution destination history information.

  According to such a configuration, the cluster system can specify the signal processing device that has executed the message processing related to the user specifying information at the time closest to the reporting time before the acquired reporting time. There is a high possibility that a cause related to the complaint from the user has occurred at the time closest to the reporting time before the acquired reporting time. Therefore, the cluster system can quickly identify a signal processing device that is highly likely to be related to a complaint from the user.

The present invention is characterized in that, when the new signal determination means selects one message from messages of the same session, the new signal is selected at the beginning of the same session.

  In the present invention, when the new signal determination unit selects one message from messages in the same session, a new message transmitted at the beginning of the same session and a subsequent message following the new message are displayed. The new message is selected based on the presence or absence of predetermined information included in the subsequent message and not included in the new message.

  According to such a configuration, the new signal determination unit can distinguish a new message and a subsequent message from messages in the same session, and can select one message from the same session. That is, the new signal determination unit can generate one log from the same session. As a result, the cluster system can reduce the number of logs of log information, so that the time required for search can be shortened, and the signal processing devices to be inspected can be quickly narrowed down.

  According to the present invention, in a scale-out type cluster system, it is possible to quickly identify a device related to a complaint such as a defect.

It is a figure which shows the structural example of the cluster system in 1st Embodiment. It is a figure which shows the example of a function of a distribution apparatus, (a) represents a main function, (b) represents an example of distribution destination information. It is a figure which shows the example of a process of a distribution destination determination part, (a) represents the case where a message is distributed among four signal processing apparatuses, (b) is a message in the state with the newly added signal processing apparatus. (C) represents a case where messages are distributed in a state where the reduced signal processing apparatus is present. It is a figure which shows the process example of a new signal determination part. It is a figure which shows the main function examples of a log | history search apparatus. It is a figure which shows an example of log information. It is a figure which shows an example of distribution destination log | history information. It is a figure which shows the example of a processing flow of a log | history search part, (a) represents the case where the signal processing apparatus before report time is specified, (b) represents the case where the signal processing apparatus nearest to report time is specified. It is a figure which shows another example of a processing flow of a log | history search part, (a) represents the case where the signal processing apparatus before report time is specified, (b) represents the case where the signal processing apparatus nearest to report time is specified. Represent. It is a figure which shows the structural example of the cluster system in 2nd Embodiment. It is a figure which shows the structural example of the cluster system in 3rd Embodiment.

  A mode for carrying out the present invention (hereinafter referred to as “the present embodiment”) will be described in detail with reference to the drawings as appropriate.

(First embodiment)
A configuration example of the cluster system in the first embodiment will be described with reference to FIG. In the first embodiment, as an example, a case will be described where SIP (Session Initiation Protocol) is used for communication and a consistent hashing method is used for a packet (message) distribution method. However, since it is not necessary to be limited to the SIP or the consistent hashing method, other implementation methods will be described.

  As shown in FIG. 1, the cluster system 100 includes a load balancer 10, a distribution device 20, a signal processing device 30, and a history search device 40. The load balancer 10 and the distribution device 20 are communicably connected via a network (second network 71). The distribution device 20 and the signal processing device 30 are communicably connected via a network (third network 72). The history search device 40 and the distribution device 20 are communicably connected by a maintenance network 73. It is assumed that there are a plurality of distribution devices 20 and a plurality of signal processing devices 30. The cluster system 100 comprises a plurality of signal processing devices 30 as a set (cluster), and flexibly changes the type (performance), number, etc. of the signal processing devices 30 in response to failures, demand fluctuations, etc. A new set shall be constructed. In FIG. 1, the second network 71 and the third network 72 are distinguished from each other, but a single network may be used.

  The load balancer 10 receives a message (for example, INVITE in SIP) from the client terminal 60 used by the user, and the received message is sent to any of the plurality of distribution devices 20 using a simple technique such as round robin. It has a function to sort and transmit.

The distribution device 20 includes a distribution destination determination unit 21 for distributing a message to an appropriate signal processing device 30 and a new signal determination unit 22 for specifying the signal processing device 30 related to a complaint such as a defect.
A distribution destination determination unit (distribution destination determination means) 21 has a function of determining a signal processing device 30 as a distribution destination using a unique value included in a message received from the load balancer 10 as a key, and transmitting the message to the distribution destination. Have Further, the distribution destination determination unit 21 has a function of transmitting configuration change information regarding the set of the signal processing device 30 to the history search device 40 in order to save the change history of the set of the signal processing device 30. Details of the processing of the distribution destination determination unit 21 will be described later.

  The new signal determination unit (new signal determination means) 22 has a function of transmitting one log to the history search device 40 for each session because messages of the same session are distributed to the same signal processing device 30. This log specifies the time at which a message is sent to the signal processing device 30 and the user who sends and receives the message in order to be able to specify the signal processing device 30 related to a complaint such as a defect from the user. Information (hereinafter referred to as user identification information) and a mapping value calculated by applying a unique value included in the message to a predetermined function. Details of the processing of the new signal determination unit 22 will be described later.

  The signal processing device 30 is, for example, a server, and has a function of generating response information for a received message and returning the response information. Since the signal processing device 30 executes complicated software processing according to the received message, the risk of occurrence of a failure or malfunction is generally higher than that of a relay system of a communication network. Therefore, the signal processing device 30 in which a failure or malfunction has occurred needs to be identified particularly quickly.

  The search unit 41 of the history search device (history search means) 40 receives the configuration change information from the distribution destination determination unit 21 and stores it in the distribution destination history information 43, receives the log from the new signal determination unit 22, and stores it in the log information 42. It has a function to memorize. Further, in order to respond to a complaint from the user, the search unit 41 receives input of information related to the complaint from the maintenance terminal 50, refers to the log information 42 and the distribution destination history information 43, and relates to the complaint. A function of specifying the signal processing device 30 to be used. Details of the function of the history search device 40 will be described later.

In FIG. 1, a client terminal 60 is a device operated by a user (for example, a personal computer or the like), and is communicably connected to the load balancer 10 of the cluster system 100 via a network (first network 70). Yes. The client terminal 60 transmits a message to the cluster system 100 and receives response information corresponding to the message from the cluster system 100.
The maintenance terminal 50 is a device (for example, a personal computer) operated by a maintenance person of the cluster system 100. The maintenance terminal 50 is communicably connected to the history search device 40 via the maintenance network 73. The maintenance terminal 50 transmits information related to the complaint to the history search device 40 and acquires the specified signal processing device 30 as response information.

Next, a function example of the sorting device 20 will be described with reference to FIG.
As shown in FIG. 2A, the sorting device 20 includes a processing unit 20a, a storage unit (storage unit) 20b, and a communication unit 20c. The processing unit 20a is configured by a CPU (Central Processing Unit) and a main memory (not shown). The application program stored in the storage unit 20b is expanded in the main memory, and the distribution destination determination unit 21 and the new signal determination unit 22 are displayed. Embodied as a function.

  The distribution destination determination unit 21 extracts a unique value included in the message received from the load balancer 10 and calculates the mapping value by applying the unique value to a predetermined function. Then, the distribution destination determination unit 21 determines a distribution destination with reference to the distribution destination information 23 (see FIG. 2B) using the calculated mapping value, and transmits the message to the distribution destination. Further, when the set of the signal processing device 30 is changed, the distribution destination determining unit 21 is a signal processing device that indicates the time when the set is changed and information for identifying the signal processing device 30 that configures the changed set. The history search device uses the ID 231 (see FIGS. 2B and 7) and the mapping value (indicated as the mapping value range 232 in FIG. 2B) associated with the signal processing device ID 231 as configuration change information. 40. In FIG. 7, the time when the set is changed corresponds to the start time of the distribution time range 431, and the time immediately before the start time corresponds to the end time of the distribution time range 431 of the set before the change.

  FIG. 2B shows an example of the distribution destination information 23. The distribution destination information 23 stores a correspondence relationship between the signal processing device ID 231 and the mapping value (denoted as a mapping value range 232 in the figure). For example, if the mapping value is 20, since it exists in “1-24” of the mapping value range 232, the signal processing device ID 231 “B” is the distribution destination. In FIG. 2B, the mapping values are not represented as individual numerical values but as mapping value ranges 232, that is, continuous numerical ranges. For example, when the mapping value associated with the signal processing device ID 231 is a skipped numerical value, the association with the signal processing device ID 231 is expressed using individual mapping values instead of the mapping value range 232. It doesn't matter.

Here, as a processing example of the distribution destination determination unit 21, a case where the consistent hashing method is used to determine the distribution destination will be described with reference to FIG.
FIG. 3A shows a state in which four signal processing apparatuses A, B, C, and D (30) are mapped to the ID space (in FIG. 3, an integer of 0 to 119). This mapping is performed, for example, by plotting a value calculated by applying a unique identifier (IP address, host name, etc.) of each signal processing device 30 to a hash function on an ID space (solid circle). In FIG. 3, it is assumed that the ID space values of the signal processing devices A, B, C, and D (30) are 0 (120), 24, 63, and 95, respectively. Now, an ID space value (mapping value) is calculated by applying the unique value in the received message to a hash function that outputs the ID space, and is mapped to the ID space (0 to 119). For example, when the mapping value (ID space value) of the message is 35, the signal processing device C (30) of “ID space value = 65” that is mapped most recently with the mapping value “35” or more is It is determined as a distribution destination. Since the ID space value circulates, the next value of 119 is considered to be 0 (120).

  In other words, the description of the mapping described above is applied to the processing of the distribution destination determination unit 21. In other words, the distribution destination determination unit 21 applies the unique value included in the received message to a predetermined function (hash function) to determine the mapping value ( ID space value) is calculated, and the distribution destination signal processing device 30 is determined with reference to the distribution destination information 23 (see FIG. 2B) using the calculated mapping value. Then, when the mapping value of the message is 35, the distribution destination determination unit 21 refers to the distribution destination information 23, and the mapping value “35” has the mapping value range 232 in “25-63”. Therefore, the distribution destination is determined to be the signal processing device C (30). Further, when the set of the signal processing device 30 is changed, the distribution destination determination unit 21 updates the distribution destination information 23 so as to correspond to the changed set, and determines the distribution destination.

  Further, by using a value unique to a session established by a message (INVITE in SIP) as a unique value in the message, it is possible to distribute all messages in the same session to the same signal processing device 30. As a value unique to the session, for example, a character string obtained by combining a SIP Call-ID and a From tag can be used. However, it is conceivable to use only information relating to the calling user who sent the message as a value unique to the session. However, if a specific user sends a large number of messages, a specific signal processing device 30 is used. Since the load is biased, it is not preferable from the viewpoint of load distribution.

  FIG. 3B shows a case where a signal processing device E (30) is newly added to the state of FIG. Here, if the ID space value of the signal processing device E (30) is 45, since the mapping value of the received message is 35, the “ID” mapped most recently with the mapping value “35” or more. The signal processing device E (30) having the space value = 45 is determined as the distribution destination.

  FIG. 3C shows a case where the signal processing devices C and E (30) are subtracted from the state of FIG. In this case, since the mapping value of the received message is 35, the signal processing device D (30) of “ID space value = 95” that is mapped most recently with the mapping value “35” or more is assigned to the distribution destination. To be determined.

  Returning to FIG. 2, when the set of the signal processing device 30 is changed, the distribution destination determining unit 21 can change the signal processing device 30 of the distribution destination related to the complaint from the user. The signal processing device ID 231 (see FIG. 2 (b)) indicating the time when the set was changed, the information for identifying the signal processing device 30 constituting the changed set, and the signal processing device ID. The mapping value is transmitted to the history search device 40 as configuration change information. Then, the history search device 40 stores the received configuration change information in the distribution destination history information 43 (see FIG. 7).

Next, a processing example of the new signal determination unit 22 will be described with reference to FIG.
The new signal determination unit 22 executes processing for extracting a new message from messages in the same session, and transmits a log of the new message to the history search device 40. The new message is a signal transmitted first in the same session (denoted as the same session in FIG. 4) as shown in FIG. The reason for extracting a new message is that a message for the same session is transmitted to the same signal processing device 30, so that one log is sufficient for each session when specifying the signal processing device 30 related to a complaint such as a defect. It is. In addition, extracting one message (here, a new message) for each session is useful for reducing the number of logs, and also has an effect of reducing the load on the sorting apparatus 20 for transmitting logs. However, in the first embodiment, the load balancer 10 distributes and transmits a message to one of the plurality of distribution devices 20 using a simple method such as round robin, so messages in the same session are transmitted to the same distribution device 20. It is not always transmitted.

  Therefore, as shown in FIG. 4, in the same session, a new message (INVITE or the like) is transmitted first, and subsequent messages (ACK, BYE, etc.) are transmitted and received thereafter. Specifically, the signal processing device 30 generates session identification information when a new message arrives, and adds the session identification information to the response information. The subsequent message carries session identification information. Therefore, the new signal determination unit 22 can determine a signal to which session identification information (predetermined information) is not assigned as a new message.

As another example, depending on the provided service or protocol, the new signal determination unit 22 determines a specific type of message (INVITE) transmitted first in the same session as a new message, and other types of messages. It is also possible to determine (UPDATE, etc.) as a subsequent message, and to distinguish a new message according to the message type.
As described above, the new signal determination unit 22 distinguishes a new message from a subsequent message.

  Then, as shown in FIG. 4, when the new signal determination unit 22 determines that the received message is a new message (INVITE), the new signal determination unit 22 generates a log and searches the log for history when distributing and transmitting the message. Transmit to device 40. And a log is memorize | stored in the log information 42 (refer FIG. 2, FIG. 6) by the search part 41 (refer FIG. 2) of the log | history search apparatus 40. FIG.

  The new signal determination unit 22 stores a log in a buffer memory (not shown), and transmits when the CPU load of the distribution device 20 is low or transmits at a predetermined cycle to control the transmission timing May be provided. In addition, the new signal determination unit 22 controls to reduce the load for establishing a connection by matching the log transmission timing with the timing of transmitting information other than the log (for example, billing information). May be provided.

Returning to FIG. 2A, the storage unit 20b is a storage device such as a memory or a hard disk, and stores an application program and a distribution destination information 23 for realizing the functions of the processing unit 20a.
The communication unit 20c is a communication interface, and transmits / receives information to / from the load balancer 10, the signal processing device 30, and the history search device 40.

Next, a function example of the history search apparatus 40 will be described with reference to FIG.
The history search device 40 includes a processing unit 40a, a storage unit (storage unit) 40b, and a communication unit 40c. The processing unit 40a is constituted by a CPU and a main memory (not shown), and an application program stored in the storage unit 40b is developed in the main memory, and the search unit 41 is realized as a function.

  The search unit 41 receives from the maintenance terminal 50 input of information (user specifying information 422) that is included in the message received from the client terminal 60 and specifies the user who transmits and receives the message, and stores the log information 42 and the distribution in the storage unit 40b. It has a function of referring to the previous history information 43 and specifying the signal processing device 30 that has processed a message related to the user. Details of the processing flow of the search unit 41 will be described later.

  The storage unit 40b is a storage device such as a memory or a hard disk, and stores an application program for realizing the function of the processing unit 40a, log information 42, and distribution destination history information 43.

  As illustrated in FIG. 6, the log information 42 is information that associates a time 421, user identification information 422, and a mapping value 423. Time 421 represents the time at which the distribution device 20 distributes a new message and transmits it to the signal processing device 30. The user specifying information 422 represents information that can be extracted from a new message and can specify either or both of the calling user (calling terminal) and the called user (calling terminal). Specifically, the user specifying information 422 is identification information for identifying the transmitting side or the receiving side. The mapping value 423 represents a value calculated by applying a unique value included in the received new message to a predetermined function (for example, a hash function).

  For example, in the first line of FIG. 6, the time 421 “20120819000001” represents August 19, 2012 00:00:01. Further, each of the user specifying information 422 “abc1” and “abc2” represents information for specifying the calling user and the called user. The user identification information 422 may be one or three or more. The mapping value 423 “90” represents a numerical value (ID space value) calculated using a unique value included in the new message.

  As shown in FIG. 7, the distribution destination history information 43 is information in which a distribution time range 431, a signal processing device ID 231 and a mapping value range 232 are associated with each other. The distribution time range 431 represents a period during which the set of signal processing devices 30 is fixed. The signal processing device ID 231 and the mapping value range 232 are the same as the distribution destination information 23 shown in FIG. Since the change history of the message distribution destination is stored in the distribution destination history information 43, even if the set of the signal processing device 30 is changed (increase / decrease in number, change in type (performance), etc.) The processing device 30 can be specified. When the search unit 41 acquires the time when the set is changed, the search unit 41 sets the acquired time as the start time of the distribution time range 431, and sets the time immediately before the start time as the distribution time range of the set before the change. The end time of 431 is set.

  In this way, the history search device 40 stores the log in the log information 42 and stores the configuration change information in the distribution destination history information 43. When the maintenance person receives a complaint from the user such as “Could not communicate between the user X and the user Y at a certain time” from the user, the history search device 40 operates the maintenance terminal 50 operated by the maintenance person. Then, input of user specifying information for specifying the user and report time is accepted. The reporting time is a time corresponding to the time when the complaint is received or the “certain time” reported by the user. And the search part 41 of the log | history search apparatus 40 performs the process which specifies the signal processing apparatus 30 which processed the said user's message using the acquired user specific information and report time. Details of the processing will be described below with reference to FIGS.

  In step S <b> 801 of FIG. 8A, the search unit 41 acquires the reporting time and user identification information from the maintenance terminal 50. Specifically, the reporting time and the user identification information are input to the history search device 40 by the maintenance person via the maintenance terminal 50.

  In step S802, the search unit 41 refers to the log information 42 and extracts an entry (first entry) that matches the acquired user identification information. Note that an entry is a set of lines indicated in the log information 42, that is, a time 421, user identification information 422, and a mapping value 423 (that is, a log).

  In step S803, the search unit 41 refers to the distribution destination history information 43 (see FIG. 7), and the mapping value 423 included in the entry (first entry) is the mapping value stored in the distribution destination history information 43. The signal processing device ID when included in the range 232 is specified. In step S803, one or more signal processing device IDs are specified. In other words, the search unit 41 refers to the distribution destination history information 43 using the mapping value 423 included in the first entry, and the mapping value 423 included in the first entry is stored in the distribution destination history information 43. The signal processing device ID is identified when it matches the mapping value.

  Next, in FIG. 8B, when a plurality of entries (first entries) are extracted in step S802, the latest signal before the reporting time from among the extracted entries (first entries). Processing for specifying the processing device ID will be described. The reason for executing this processing is that the time closest to the reporting time before the reporting time is assumed to be most likely to have caused a cause related to the complaint from the user.

  In step S804, the search unit 41 refers to the log information 42 (see FIG. 6), and from among the plurality of entries (first entries) extracted in step S802, the time closest to the reporting time before the reporting time. Entries (second entry) are extracted. Next, the detection unit 41 refers to the distribution destination history information 43, and includes the time 421 included in the entry (second entry) included in the distribution time range 431 and the mapping included in the entry (second entry). The signal processing device ID when the value 423 is included in the mapping value range 232 stored in the distribution destination history information 43 is specified. In other words, when there are a plurality of extracted first entries, the search unit 41 extracts, from the plurality of first entries, the first entry at the time closest to the reporting time before the acquired reporting time. And the distribution destination history information 43 is referred to using the set of the time 421 and the mapping value 423 included in the second entry, and the time 421 is included in the distribution time range 431 and the mapping value The signal processing device ID when 423 matches the mapping value stored in the distribution destination history information 43 is specified.

Next, another processing flow example of the search unit 41 will be described with reference to FIGS. This different processing flow example is different from the processing flow example shown in FIG. 8A in that, in step S902, when “extracting an entry (first entry) with which the acquired user identification information matches” is extracted, The search range is limited by setting the previous predetermined period.
In step S <b> 901 of FIG. 9A, the search unit 41 acquires the reporting time and user identification information from the maintenance terminal 50. Specifically, the reporting time and the user identification information are input to the history search device 40 by the maintenance person via the maintenance terminal 50.

  In step S902, the search unit 41 refers to the log information 42 and extracts an entry (first entry) that matches the acquired user identification information 422 in a predetermined period before the acquired reporting time.

  In step S903, the search unit 41 refers to the distribution destination history information 43 (see FIG. 7), the time 421 included in the entry (first entry) is included in the distribution time range 431, and the entry (first entry). The signal processing device ID in the case where the mapping value 423 included in is included in the mapping value range 232 stored in the distribution destination history information 43 is specified. In step S903, one or more signal processing device IDs before the reporting time are specified. In other words, the search unit 41 refers to the distribution destination history information 43 using a set of the time 421 and the mapping value 423 included in the first entry, and the time 421 is included in the distribution time range 431 and the mapping is performed. The signal processing device ID when the value 423 matches the mapping value stored in the distribution destination history information 43 is specified.

  Next, in FIG. 9B, when a plurality of entries (first entries) are extracted in step S902, the latest signal before the reporting time from among the extracted entries (first entries). Processing for specifying the processing device ID will be described. The reason for executing this processing is that the time closest to the reporting time before the reporting time is assumed to be most likely to have caused a cause related to the complaint from the user.

  In step S904, the search unit 41 refers to the log information 42 (see FIG. 6), and from among the plurality of entries (first entries) extracted in step S902, the time closest to the reporting time before the reporting time. Entries (second entry) are extracted. Next, the search unit 41 includes the time 421 included in the entry (second entry) in the distribution time range 431 and the mapping value 423 included in the entry (second entry) in the distribution destination history information 43. The signal processing device ID when it is included in the stored mapping value range 232 is specified. In other words, when there are a plurality of extracted first entries, the search unit 41 extracts, from the plurality of first entries, the first entry at the time closest to the reporting time before the acquired reporting time. Referring to the distribution destination history information 43 using the combination of the time 421 and the mapping value 423 included in the second entry, the time 421 is included in the distribution time range 431 and the mapping value 423 Identifies the signal processing device ID when the mapping value matches the mapping value stored in the distribution destination history information 43.

  As described above, in the processing flow examples shown in FIGS. 8 and 9, the cluster system 100 can specify one or more signal processing devices 30 related to the user's complaint as inspection targets in steps S803 and S903. Further, the cluster system 100 can specify one signal processing device 30 related to the user at the time closest to the time before the acquired time as the inspection target by steps S804 and S904. There is a high possibility that a cause related to the complaint from the user has occurred at the time closest to the time before the acquired time. Therefore, the cluster system can quickly identify a signal processing device that is highly likely to be related to a complaint from the user. If the processing flows shown in FIGS. 8 and 9 are not executed, all the signal processing devices 30 constituting the cluster system 100 are to be inspected, which requires a lot of time. On the other hand, in the first embodiment, the signal processing device 30 to be inspected can be quickly identified, so that it is possible to quickly confirm the state of the defect, investigate the cause, and recover.

  In addition, it has been described that the first entry is extracted in steps S803 and S903, and the second entry is extracted from the first entry in the next steps S804 and S904. However, as a modification, in the steps S802 and S902, The second entry may be extracted.

  In the first embodiment, by providing the history search device 40 separately from the distribution device 20 and the signal processing device 30, the distribution device 20 and the signal processing device 30 are subjected to a load of search processing different from the original processing. Can not be.

(Second Embodiment)
As a modification of the cluster system 100 of the first embodiment, a configuration example of the cluster system 100A in the second embodiment will be described with reference to FIG. The same block configuration and function as those in FIG. 1 are denoted by the same reference numerals and description thereof is omitted.
The cluster system 100A of the second embodiment is different from the cluster system 100 of the first embodiment in that the distribution device 20A includes a distribution destination determination unit 21, the signal processing device 30A includes a new signal determination unit 22A, and a maintenance network. That is, the signal processing device 30 </ b> A and the history search device 40 are communicably connected via the network 73.

Unlike the new signal determination unit 22 of the first embodiment, the new signal determination unit 22A of the signal processing device 30A can prevent a log of the retransmission message from being output when a retransmission message is received.
Specifically, the case of SIP will be described as an example. The message includes (To, To tag, From, From tag, Call-ID), and the value of CSeq is incremented each time a message is generated. In the case of resending, the already received message (To, To tag, From, From tag, Call-ID) is the same, and the value of CSeq is also set to the same. Therefore, it is possible to distinguish between a newly generated message and a retransmitted message.

  That is, in the second embodiment, since the newly generated message and the retransmit message are distributed to the same signal processing device 30A, the signal processing device 30A does not output a log when it is determined that the message is a retransmit message. can do. As a result, the number of logs can be reduced, and the search time by the search unit 41 can be shortened. The other effects in the second embodiment are the same as those in the first embodiment.

  However, in the second embodiment, when the signal processing device 30A is removed (detached) from the set, the history search device 40 always stores all logs before the signal processing device 30A is removed (detached). It is preferable that the information has already been transmitted.

(Third embodiment)
As a modification of the cluster system 100 of the first embodiment, a configuration example of the cluster system 100B in the third embodiment will be described with reference to FIG. The same block configuration and function as those in FIG. 1 are denoted by the same reference numerals and description thereof is omitted.
The cluster system 100B of the third embodiment is different from the cluster system 100 of the first embodiment in that the distribution device 20B includes history search means 40B that is a function of the history search device 40, and the history search device 40 is provided independently. It is not.

  The history search unit 40B of the distribution device 20B includes the search unit 41, log information 42, and distribution destination history information 43 described in FIG. 1 of the first embodiment. However, the history search means 40B obtains the configuration change information from the distribution destination determination unit 21 of the distribution device 20B provided therein, stores it in the distribution destination history information 43, and determines a new signal of the distribution device 20B provided therein. A log is acquired from the unit 22 and stored in the log information 42. Therefore, when the signal processing device 30 that may be related to the claim is specified, the maintenance terminal 50 accesses all the history search means 40B of the sorting device 20B, and the signal processing device specified by the history search means 40B. Get an ID.

  The third embodiment differs from the first embodiment in that the distribution device 20B does not have to send logs and configuration change information to the maintenance network 73, so that the communication processing load can be reduced. What is the network failure of the maintenance network 73? Logs and configuration change information can be stored independently. The other effects in the third embodiment are the same as those in the first embodiment.

  As described above, the cluster system (100, 100A, 100B) of each embodiment includes the load balancer 10, the distribution device 20, the signal processing device 30, and the history search device 40 (history search means 40B).

  The distribution destination determination unit 21 transmits configuration change information of the signal processing device 30 to the history search device 40 when the set of the signal processing device 30 is changed. Further, the new signal determination unit 22 determines a new message in the same session and transmits a log of the new message to the history search device 40. The search unit 41 of the history search device 40 stores the log received from the distribution device 20 in the log information 42 (see FIG. 6), and stores the configuration change information in the distribution destination history information 43 (see FIG. 7). Since the history search device 40 includes the distribution destination history information 43 (see FIG. 7), even if the set of the signal processing device 30 is changed during the same session, message distribution is performed. The previous signal processing device 30 can be tracked.

  Then, the search unit 41 acquires the reporting time and user identification information from the maintenance terminal 50, and refers to the log information 42 and the distribution destination history information 43, and is related to the user identification information acquired before the reporting time. The signal processing device ID is specified.

  In the first embodiment, as shown in FIG. 4, since the load balancer 10 distributes to one of the plurality of distribution devices 20 using a simple method such as round robin, messages in the same session are different. A method for determining a new message in the case of being transmitted to the server has been described. Even when messages of the same session are transmitted to the same sorting apparatus 20, the same method as the above-described new message determination method can be used.

  Although not shown, the history search device 40 may be provided in the maintenance terminal 50. In this case, the function of the cluster system 100 transmits logs and configuration change information to the maintenance terminal 50. Then, the maintenance person can easily perform the operation of specifying the signal processing device 30 to be inspected with reference to the log information 42 and the distribution destination history information 43 in the maintenance terminal 50.

DESCRIPTION OF SYMBOLS 10 Load balancer 20, 20A, 20B Sorting apparatus 20a Processing part 20b Storage part (storage means)
20c communication unit 21 distribution destination determination unit (distribution destination determination means)
22, 22A New signal determination unit (new signal determination means)
23 Distribution destination information 30, 30A Signal processing device 40 History search device (history search means)
40B History search means 40a Processing section 40b Storage section (storage means)
40c communication unit 41 search unit 42 log information 43 distribution destination history information 50 maintenance terminal (terminal)
60 Client terminal 100, 100A, 100B Cluster system 231 Signal processing device ID
232 Mapping value range 421 Time 422 User identification information 423 Mapping value 431 Distribution time range

Claims (6)

A cluster system comprising a set composed of a plurality of signal processing devices,
One message is extracted from messages in the same session, and the user identification information indicating the information specifying the user who transmits and receives the message included in the extracted message and the unique value included in the message are applied to a predetermined function. New signal determining means for transmitting the calculated mapping value as a log to the history detecting means,
When the set of signal processing devices is changed, the signal processing device ID indicating information for identifying the signal processing devices constituting the changed set and the mapping value associated with the signal processing device ID are changed. A distribution destination determination unit that transmits information to the history detection unit;
The log is received and stored in the log information of the storage means, the configuration change information is received and stored in the distribution destination history information of the storage means,
The user identification information input from the terminal is acquired, the log information is referred to, the first entry indicating the log that matches the acquired user identification information is extracted, and the mapping value included in the first entry A history search for identifying the signal processing device ID when the mapping value included in the first entry matches the mapping value stored in the distribution destination history information with reference to the distribution destination history information using Means,
A cluster system comprising: The new signal determination means further transmits the extracted message including the time when the extracted message is transmitted,
The distribution destination determining means further transmits the configuration change information including the time when the set was changed,
The history search means, when there are a plurality of the extracted first entries, indicates the time range in which the set was configured based on the time when the set was changed after receiving the configuration change information A distribution time range is generated and stored in the distribution destination history information in association with a signal processing device ID indicating information for identifying a signal processing device constituting the set, and a reporting time is acquired from the terminal, From the first entry, the first entry at the time closest to the reporting time before the acquired reporting time is extracted as a second entry, and a set of time and mapping value included in the second entry is used. Referring to the distribution destination history information, the time is included in the distribution time range, and the mapping value is stored in the distribution destination history information Cluster system of claim 1, wherein said signal processing device ID to identify when matching the serial mapping values. A cluster system comprising a set composed of a plurality of signal processing devices,
One message is extracted from messages in the same session, the time when the extracted message is transmitted, user identification information indicating information included in the message and identifying a user who transmits and receives the message, and unique information included in the message A new signal determination unit that transmits a mapping value calculated by applying a value to a predetermined function to the history detection unit as a log;
When the set of signal processing devices is changed, the time at which the set is changed, the signal processing device ID indicating information for identifying the signal processing devices constituting the changed set, and the signal processing device ID A distribution destination determination means for transmitting the attached mapping value to the history detection means as configuration change information;
The log is received and stored in the log information of the storage means, and the configuration change information is received and the distribution time range indicating the time range in which the set was configured is generated based on the time when the set was changed Then, the distribution destination history information of the storage means in association with the signal processing device ID indicating information for identifying the distribution processing time range and the signal processing device constituting the set, and the mapping value associated with the signal processing device ID And remember
The reporting time input from the terminal and the user identification information are acquired, and the log information is referred to to indicate the log in which the acquired user identification information matches within a predetermined period before the acquired reporting time. The first entry is extracted, the distribution destination history information is referred to using the set of time and mapping value included in the first entry, the time is included in the distribution time range, and the mapping value is History search means for specifying the signal processing device ID when matching with the mapping value stored in the distribution destination history information;
A cluster system comprising: In the case where there are a plurality of the extracted first entries, the history search means selects the first entry at a time closest to the reporting time before the acquired reporting time from the plurality of first entries. The second entry is extracted, and the distribution destination history information is referred to using the combination of the time and mapping value included in the second entry, the time is included in the distribution time range, and the mapping value is The cluster system according to claim 3, wherein the signal processing device ID is identified when the mapping value matches the mapping value stored in the distribution destination history information. The said new signal determination means selects the new message transmitted at the beginning of the said same session, when selecting one message from the messages of the same session, Any one of Claim 1 thru | or 4 characterized by the above-mentioned. Or the cluster system according to one item. When the new signal determination unit selects one message from messages in the same session, the new message is included in the subsequent message including a new message transmitted first in the same session and a subsequent message following the new message. 5. The cluster system according to claim 1, wherein the new message is selected based on the presence or absence of predetermined information not included in the new message.

Images