JP5412481B2 - Data synchronization method - Google Patents

Description

  The present invention relates to a data synchronization method in a duplex system, and more particularly to a data synchronization method in a call control system that controls session establishment by call control.

  2. Description of the Related Art Conventionally, it has been known that a system is duplicated as one of methods for improving system reliability (see, for example, Patent Document 1). A system that performs processing during normal operation is called an active system, and a system that takes over when a failure occurs in an active system is called a standby system. When a failure occurs, the active system and the standby system are switched, and thereafter, processing is performed on the new active system.

JP 2010-28708 A

  By the way, in the original development of each communication company based on the conventional exchange, it takes a long development period and the development cost is high, so a service availability forum (SAF) was established by a company related to the communication business and its device development. SAF aims to increase portability and reduce costs by standardizing specifications. In recent years, devices installed in a plurality of IP networks cooperate with each other to provide a complicated service, and therefore, an object is also to improve interoperability.

  In SAF, specifications are determined according to several classifications. Among them, there is a checkpoint (CKPT) mechanism in which data is copied (synchronized) between nodes so that service can be continued using previous data when a failure occurs. That is, when a failure occurs, the active system and the standby system are switched, and call processing is performed using data synchronized by the CKPT function in the new active system. At this time, call processing has already started in the new active system, the received signal of the new call is stored as data, and editing is started. Therefore, since the new active system is in the latest state with reliability, the CKPT data possessed by the new active system is copied (re-synchronized) from the new active system to the new standby system. As a result, it is possible to resume (continue) the call processing from the state before the occurrence of the failure.

  However, if the call information of the new working system is updated (changed) during the resynchronization of the CKPT data, an information conflict occurs between the new working system and the new standby system. In order to avoid such a contradiction, it is necessary to manage the synchronization / unsynchronization for the change or the synchronization / unsynchronization management for the change. In usages where CKPT is frequently used, the updated location may be further updated. In such a case, management for the updated part becomes necessary, and the processing becomes further complicated. For this reason, conventionally, the call processing is resumed after the resynchronization of the CKPT data is completed. Or, it was impossible to continue the call due to loss of call information.

  An object of the present invention is to provide a data synchronization method capable of resuming (continuing) call processing promptly after system switching in view of the above-described conventional technology.

To achieve the above object, according to the first aspect of the invention, call processing during normal operation is performed by the active call control server, and when a failure occurs, the active call control server and the standby call control server are switched. A data synchronization method on the new active call control server in a call control system that performs call processing on the new active call control server, and resynchronizes data in the basic area of the new active call control server after system switching. The first synchronization step started from the new active call control server to the new standby call control server and the data generated by the received signal including the call processing continuation signal are stored in the temporary area of the new active call control server. A saving step for saving, and a second synchronization step for starting the synchronization of the data saved in the temporary area of the new active call control server from the new active call control server to the new standby call control server. If the call processing control unit for controlling the call processing receives a write request to the temporary area, by delaying its response to control by using the temporary area throughput of the new active system call control server The gist.

The invention according to the second aspect is characterized in that, in the invention according to the first aspect, the delay time of the response is set to (minimum retransmission interval−internal signal processing time).

The invention according to a third aspect is the invention according to the first aspect, wherein the write amount from the temporary area to the basic area is less than the write amount from the temporary area to the basic area. The gist is to reduce it.

  According to the present invention, it is possible to provide a data synchronization method capable of resuming (continuing) call processing immediately after system switching and reducing the load caused by resuming (continuing) call processing. is there.

It is a network block diagram of the SIP communication system in Embodiment 1 of this invention. It is explanatory drawing of CKPT in Embodiment 1 of this invention. It is a functional block diagram of the call control system in Embodiment 1 of the present invention. It is a flowchart which shows operation | movement of the call control system in Embodiment 1 of this invention. It is explanatory drawing of the data synchronization method in Embodiment 1 of this invention. It is a functional block diagram of the new ACT system call control server in Embodiment 2 of this invention. It is a flowchart which shows operation | movement of the call control system in Embodiment 2 of this invention.

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

(Embodiment 1)
FIG. 1 is a network configuration diagram of a SIP communication system 1 according to the first embodiment. The SIP communication system 1 is a system that performs call control by SIP (Session Initiation Protocol). As shown in FIG. 1, subscriber terminals 2A, 2B, 2C,. Transfer devices 3A, 3B, 3C,..., Subscriber accommodation servers 4A, 4B, relay servers 5A, 5B, a call control system 6, and guidance servers 7A, 7B.

  The subscriber terminals 2A, 2B, 2C,... Are IP telephones, HGWs (home gateways), etc. used by subscribers who are users, and a function of performing SIP negotiation for media communication used between the subscriber terminals. Have The transfer devices 3A, 3B, 3C,... Transfer the media communication traffic of the subscriber terminals 2A, 2B, 2C,. The subscriber accommodation server 4A is a proxy server that relays SIP signals of the subscriber terminals 2A, 2B, and 2C, and the subscriber accommodation server 4B is a proxy server that relays SIP signals of the subscriber terminals 2D, 2E, and 2F. is there. The subscriber accommodation servers 4A and 4B have a function of relaying the SIP negotiation between the subscriber terminals, supplementing the SIP signal based on the subscriber contract information, and a function of editing the SIP signal according to a rule stored in advance. The relay servers 5A and 5B have a function of connecting the subscriber accommodation servers 4A and 4B and relaying SIP signals. The call control system 6 includes an ACT (active) call control server 61 and an SBY (standby) call control server 62.

  The subscriber accommodation servers 4A and 4B perform services using information of subscribers within the accommodation range. On the other hand, the ACT call control server 61 and the SBY call control server 62 provide services that span a plurality of subscriber accommodation servers, such as nationwide services. Typical examples include a number conversion service such as toll free and a guidance service that sends guidance according to the status of the incoming user. The way of providing the service is as follows.

  First, when a user specifies a logical telephone number from the subscriber terminal 2A and makes a call, a call signal such as an invite signal is branched from the transfer device 3A at the edge of the IP network 3 and sent to the subscriber accommodation server 4A. Sent. After the processing based on the subscriber information is performed, if the destination is a logical number, it is sent to the call control system 6 and converted into a physical number as the destination. Thereafter, it is returned to the original subscriber accommodation server 4A or connected to the subscriber accommodation server 4B that accommodates the physical number that is the destination. Depending on the network configuration and number band, it is connected to the destination subscriber accommodation server 4B via the relay servers 5A and 5B. The calling subscriber terminal 2A is called the calling side, and the called subscriber terminal 2D is called the called side. Acceptable conditions are exchanged between the subscriber terminals 2A and 2D (SIP negotiation), and media communication is performed based on the established conditions. The media communication here is generally a call. A signal (packet) used for media communication is delivered via the access network 2 and the IP network 3. It does not go through the subscriber accommodation servers 4A and 4B. Guidance voice of guidance communication is played via the guidance servers 7A and 7B. Since it is intended for nationwide calls and signals are transmitted simultaneously from a plurality of subscriber accommodation servers, the processing frequency is extremely high. There is a need for high availability that can provide these high-frequency call processing without interruption.

  When the call control system 6 receives the call signal from the calling party accommodating server 4A, the call control system 6 performs call processing while performing call processing such as setting of the destination IP address and timer and changing the route information in addition to number conversion. change. Then, as shown in FIG. 2, the call processing is performed while copying (synchronizing) the CKPT data from the ACT call control server 61 to the SBY call control server 62, and the processing result is sent to the originating subscriber accommodation server. Respond to 4A. Alternatively, the information may be transmitted to the called-side subscriber accommodation server 4B or may be instructed to the guidance servers 7A and 7B. In some cases, the originating side subscriber accommodation server 4A and the destination side subscriber accommodation server 4B are the same. As for the unit of synchronization, each developer is given discretion within the range of CKPT maximum size (= number of sections × section size). Here, it is assumed that the call from the start of arrival / departure communication to the end of communication is synchronized every time it is updated using one section. The communication start may be made not only from the caller but also from the call control system 6. The end of communication may be disconnected unintentionally such as communication disconnection. Including these cases, one section is used for calls from the start of communication to the end of communication. Since signals that can be expected to be retransmitted can be re-executed by retransmission even if data is lost, they are synchronized at the timing of transmission from the own server.

  FIG. 3 is a functional block diagram of the call control system 6. FIG. 3A shows before system switching, and FIG. 3B shows after system switching. The call processing control units 611 and 631 are processes that control call processing and operate in the ACT system. The synchronization processing control unit 612 and the like are processes that control synchronization processing with other systems. The synchronization area storage unit 613 is specifically a CKPT buffer.

  As shown in FIG. 3A, call processing during normal operation is performed by the ACT call control server 61. When a failure occurs, the ACT call control server 61 and the SBY call control server 62 are switched. As a result, the old SBY call control server 62 becomes the new ACT call control server 63 and the old ACT call control server 61 becomes the new SBY call control server 64, as shown in FIG. The new ACT call control server 63 continues (resumes) the call processing. In this case, the route through which the data passes is in the order of the call processing control unit 631, the temporary area 63A, the basic area 63A, the temporary area 64A, and the basic area 64B. By the way, after the system switching, it is necessary to continue (restart) the call processing in the new ACT call control server 63 promptly. The new ACT call control server 63 can continue the call processing for a continuous call using the synchronized CKPT data. At this time, since the new ACT call control server 63 is in the latest reliable state, it is necessary to resynchronize the CKPT data from the new ACT call control server 63 to the new SBY call control server 64.

  However, if the call information of the new ACT call control server 63 is updated during the resynchronization of the CKPT data, information conflict occurs between the new ACT call control server 63 and the new SBY call control server 64. . Therefore, conventionally, the call processing is resumed after the resynchronization with the CKPT data is completed. Or, it was impossible to continue the call due to loss of call information. In the present embodiment, the following method is employed in order to be able to resume (continue) call processing immediately after system switching.

  FIG. 4 is a flowchart showing the operation of the call control system 6. Here, a description will be given focusing on the data synchronization method on the new ACT call control server 63.

  First, when a system change occurs due to a failure, resynchronization of data in the basic area 63B of the new ACT call control server 63 is started from the new ACT call control server 63 to the new SBY call control server 64 (step). S11). The basic area 63B is a storage area in the CKPT and corresponds to a plurality of sections shown in FIG. Next, the data generated by the received signal including the call processing continuation signal is saved in the temporary area 63 A of the new ACT call control server 63. Here, “evacuation” means temporarily storing. At this time, since the new ACT call control server 63 is in a high load state by resuming the call processing, the processing amount of the new ACT call control server 63 during resynchronization is controlled so as to reduce this load. ing. Specifically, as shown in FIG. 3B, when the synchronous processing control unit 632 of the new ACT call control server 63 receives a write request to the temporary area 63A from the call processing control unit 631, the response Is delayed by a predetermined time (steps S12 → S13 → S14 → S15).

  In the main flow of call processing, an SIP signal is received and a response is returned after the signal is stored and edited and written to another system. The main flow of call processing is performed by a plurality of worker threads. Each worker thread is carried out until a response is returned, and no other call processing is performed. Therefore, by delaying the response by CKPT management, which is one process in the worker thread, it becomes possible to easily adjust the processing amount to the desired CKPT.

  The delay time of the write response is set to an appropriate time that does not overflow the CKPT buffer even if the write response is returned during resynchronization. In other words, the new ACT call control server 63 during resynchronization transfers the CKPT data to the new SBY call control server 64 and simultaneously processes the write request from the call processing control unit 631. Therefore, a mechanism for preventing the prepared CKPT buffer from overflowing is necessary. Thus, until the resynchronization is completed, as described above, the write request is limited by delaying the write response. Thereby, the time required for the call processing control unit 631 becomes longer than the interval of calls accepted by the call control system 6. However, increasing the call interval is necessary to reduce the call volume per unit time processed by the call control system 6 and to operate normally within the limited CKPT resource during resynchronization.

  Hereinafter, the delay time will be described in more detail.

  For example, when there is a call volume of 300,000 BHCA (Busy Hour Call Attempts), the processing request frequency is every 3600000 ms ÷ 300000 = 12 ms. BHCA is the total number of line calls in the time zone when the telephone network is most congested. If the signal processing is delayed too much, conversely, the amount of signal processing by the retransmission signal doubles from the previous stage of the main flow of call processing, that is, from the stage of signal reception processing. If the delay time is less than 500 ms for the first retransmission interval of the SIP signal, retransmission does not occur. If the normal processing time (internal signal processing time) from when a SIP signal enters the call processing server to when it comes out is 50 ms, the delay time can be extended to 450 ms. The signal processing every 450 ms is equivalent to the case where the processing amount of CKPT is 8000 BHCA when calculated simply, but actually, the call amount is lowered by extending the following average holding time. In addition, since the amount of signal from the outside (the amount of signal entering the call processing server as a SIP signal) does not decrease that much, more buffers such as a signal reception buffer are required in the previous stage of call processing. Increasing the internal signal processing time increases the average hold time. Even when the average hold time is extended, the call volume is reduced due to the simultaneous connection limit as shown in the following equation. Assuming that the number of SIP signals per call is 13, each average hold time can be extended by α = 13 × 400 = 5200 ms≈5 seconds when each is delayed by 400 ms.

The number of simultaneous connections × 3600 ÷ (average hold time + α) = reduced call volume Therefore, in this embodiment, the delay time is (minimum retransmission interval−internal signal processing time). In addition, assuming that the average hold time is (minimum retransmission interval-internal signal processing time) x number of signals, the number of simultaneous connections x 3600 / (average hold time + α) = reduced call volume, extra for the reduced call volume In addition, many internal buffers such as a signal receiving buffer are designed.

  Hereinafter, such a data synchronization method will be described in more detail with reference to FIG.

  First, as shown in FIG. 5A, call processing during normal operation is performed by the ACT call control server 61, and writing to the CKPT of the other system 62 is waited for each call, and the write response is returned. Different calls use different sections as shown in FIG. 5 (b). When system switching occurs due to a failure, the new ACT system call control server 63 performs call processing as shown in FIG. The new ACT call control server 63 resynchronizes the data in the basic area 63B with the new SBY call control server 64. Therefore, a new call (circle in the figure) is written in the temporary area 63A of the own system 63 during resynchronization, but a write response is returned because a predetermined time has not passed at the time of FIGS. 5 (c) and 5 (d). Absent. When a predetermined time elapses for each call, as shown in FIG. 5E, the synchronization of the data saved in the temporary area 63A is started, and writing to the CKPT of the other system 64 is waited for each call. Returns a write response. Specifically, the continuation call (dotted line) or new call (dashed line) saved in the temporary area 63A is transferred to the new SBY call control server 64 while being written to the basic area 63B of the new ACT call control server 63. It has become. When the resynchronization is completed, as shown in FIG. 5F, the process returns to the call processing during normal operation.

  As shown in FIG. 5, call processing is performed using the basic area 63B from the normal operation via the temporary area 63A (or while determining whether it is time to pass through the temporary area 63A). This eliminates the need to separate the route that uses the temporary area 63A and the route that uses the basic area 63B, so that the processing program can be shared.

  As described above, according to the call control system 6 in the present embodiment, it is possible to provide a data synchronization method capable of resuming (continuing) call processing promptly after system switching. That is, since the new ACT call control server 63 uses the temporary area, the new SBY call control server 64 can be updated promptly while continuing the call processing without performing a new write to the basic area. Can do. In a series of call processing from utterance to end (from invite to bye), there are a number of SIP signal transmissions and receptions. Even if the system is switched, this series of call processing can be continued, so that a call during a call cannot be disconnected. A signal that performs a series of call processing from the utterance to the end is called a continuation signal, and includes an update, a re-invite, and a retransmission signal.

  In this embodiment, the processing amount of the new ACT call control server 63 during resynchronization is controlled using a temporary area so as to reduce the load caused by resumption (continuation) of call processing. Specifically, when a write request to the temporary area 63A is received from the call processing control unit 631, the response is delayed. As a result, it is possible to adjust so that the call volume during resynchronization does not become excessive, and it is possible to appropriately maintain the amount of storage per unit time in the temporary area 63A during resynchronization.

(Embodiment 2)
By the way, as shown in FIG. 5 (e), new calls saved in the temporary area 63A increase while data is written from the temporary area 63A to the basic area 63B during resynchronization. Therefore, the transfer rate from the new ACT call control server 63 to the new SBY call control server 64 increases, which may affect other transfer processes. In the present embodiment, the following method is employed in order to avoid such a problem.

  FIG. 6 is a functional block diagram of the new ACT call control server 63 in the second embodiment. During the resynchronization with the new SBY call control server 64, the new ACT call control server 63 writes more data from the call processing control unit 631 to the temporary area 63A than to write data from the temporary area 63A to the basic area 63B. Is to reduce. Here, as shown in FIG. 6A, the amount of writing from the temporary area 63A to the basic area 63B is three, whereas as shown in FIG. The case where the amount of writing to the temporary area 63A is two is illustrated.

  FIG. 7 is a flowchart showing the operation of the call control system 6 in the second embodiment.

  First, when system switching occurs due to a failure, resynchronization of data in the basic area 63B is started as in the first embodiment (step S21). At this time, since the new ACT call control server 63 is in a high load state by resuming the call processing, the processing amount of the new ACT call control server 63 during resynchronization is controlled so as to reduce this load. ing. Specifically, when three are written from the temporary area 63A to the basic area 63B, two are written from the call processing control unit 631 to the temporary area 63A (steps S22 → S23 → S24). When a certain time elapses, the call processing is returned to the normal operation.

  As described above, in the call control system 6 in the present embodiment, the write amount from the call processing control unit 631 to the temporary region 63A is made smaller than the write amount from the temporary region 63A to the basic region 63B. As a result, the transfer rate from the new ACT call control server 63 to the new SBY call control server 64 is reduced, and the influence on other transfer processes can be suppressed. As a result, it is possible to bring the writing from the temporary area 63A to the basic area 63B closer to the normal call processing state while giving priority to resynchronization.

  Here, the synchronization of the data saved in the temporary area of the new ACT call control server 63 is started from the new ACT call control server 63 to the new SBY call control server 64 during the resynchronization. However, this start timing is not limited to during resynchronization. That is, it can be started after resynchronization is completed.

  In addition, although preferred embodiment was described above, this invention is not limited to the said embodiment, A various deformation | transformation is possible. For example, although the one-to-one communication between the calling side and the called side has been described as an example, the present invention can be applied to a three-party call such as a telephone conference service. In addition, calls received during resynchronization can be calculated. For example, if 20% is a four-party call, 20% of the call volume is calculated as two sections, so the temporary area can be appropriately sized. Can be set.

6 ... Call control system 61 ... Active (ACT) call control server 62 ... Standby (SBY) call control server 63 ... New active (new ACT) call control server 63A ... New active (new ACT) Call control server temporary area 63B ... New active system (new ACT system) call control server basic area 64 ... New backup system (new SBY system) call control server 631 ... Call processing control unit

Claims (3)

Call processing during normal operation is performed by the active call control server, and when a failure occurs, the call control server switches between the active call control server and the standby call control server and performs call processing at the new active call control server. A data synchronization method on the new active call control server,
A first synchronization step of starting resynchronization of data in the basic area of the new active call control server from the new active call control server to the new standby call control server after system switching;
A evacuation step for evacuating data generated by a received signal including a call processing continuation signal to a temporary area of the new active call control server;
A second synchronization step for starting the synchronization of the data saved in the temporary area of the new active call control server from the new active call control server to the new standby call control server;
If the call processing control unit for controlling the call processing receives a write request to the temporary area, by delaying its response to control by using the temporary area throughput of the new active system call control server A featured data synchronization method. Data synchronization method according to claim 1, characterized in that the - the delay time of the response (internal signal processing time minimum retransmission interval).   2. The data synchronization method according to claim 1, wherein a write amount to the temporary area is reduced from a call processing control unit that controls call processing, rather than a write amount from the temporary area to the basic area.

Images