JP2004220118A - Multiprocessor system, monitoring system, method of taking out queue from multiprocessor system, and process for table recovery at multiprocessor system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a multiprocessor system which prevents calling losses at the time of recovery, simplifies a recovery process and increases the speed of recovery in the event of failure when a plurality of processes access a shared memory, and which provides system reliability, and flexibility derived from a buffer given a large working capacity. <P>SOLUTION: In the multiprocessor system, the shared memory is provided with a queue management information table 7 which retains queue management information consisting of a plurality of information elements specifying both the magnitude and position of a resource management queue retaining call information about communication calls gained by each processor, and a memo recording part 8 (updated history information recording area) which records history information consisting of a plurality of update information elements for writing the queue management information on the queue management information table 7. <P>COPYRIGHT: (C)2004,JPO&NCIPI

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to, for example, a load distribution type multiprocessor system, and more particularly to a multiprocessor system, a monitoring system, The present invention relates to a queue retrieval method in a multiprocessor system and a table recovery processing method in a multiprocessor system.
[0002]
[Prior art]
2. Description of the Related Art Various communication apparatuses in a W-CDMA (Wideband-Code Division Multiple Access) type mobile communication system perform transmission / reception processing of a large amount of data.
(P1) An example of a mobile communication system
A radio network controller (RNC) provided between a plurality of radio base station apparatuses (BTS: Base Transmitting Station: hereinafter referred to as BTS unless otherwise specified) and a public network or the Internet. Unless otherwise specified, the RNC receives large-capacity moving image data and the like from a public network and transfers the moving image data and the like to one or more BTSs. Also, the RNC transmits information data transmitted from a plurality of BTSs to a public network. That is, the RNC requires a large amount of call processing (Call Processing: CP) and a large amount of arithmetic processing.
[0003]
(P2) Load distribution device
For this reason, the RNC constitutes a multiprocessor system including a plurality of processors for performing signal processing and a shared memory for holding data necessary for call processing shared by the plurality of processors. Load is distributed (distributed).
[0004]
Various multiprocessor systems for performing this load distribution processing have been conventionally proposed (for example, see Patent Document 1).
The RNC queue management method (queue management method or queue management information method) manages resources for each device (or each device). For example, one resource is assigned to one BTS.
[0005]
(P3) Resources
Here, the resource means a physical memory area or a memory area holding one piece of call information such as identification information of a communication partner unless otherwise specified. The shared memory is provided in a communication device accessed (read / written) by a plurality of processors, and uses a bidirectional queue.
[0006]
(P4) Queue
A queue generally has information data and another queue address to which the queue itself connects. Here, the two-way queue includes, for example, a queue address immediately before the queue itself (hereinafter, a previous queue address) and a queue address immediately after the queue itself, such as resources A to C shown in FIG. (Hereinafter, the next queue address). Thereby, the processor can search for a desired queue by referring to each “next queue address” of the resources A to C shown in FIG. Further, the processor can refer to the “previous queue address” stored in the “next queue address” of the resource C and search each queue in the reverse direction in order. Unlike a one-way queue having only a "queue address".
[0007]
Then, the processor inserts various types of call information into the queue, and another processor refers to the call information in the queue, whereby the call information is shared between the processors via the queue. This queue is constantly managed by the queue management information table while each processor is processing call information.
(P5) Load sharing type multiprocessor system
A shared memory in an RNC system will be described as an example of a load distribution type multiprocessor with reference to FIGS.
[0008]
FIG. 19 is a block diagram of a load distribution type multiprocessor system. The load distribution device 200 shown in FIG. 19 is provided in the RNC and performs, for example, call processing. The load distribution control unit (control unit) 201, the processors A to D, the shared memory 203, and the inter-processor communication device 204 It is configured with the following.
Here, the control unit 201 receives information data and control data from the exchange 102 and the BTSs 198a to 198b, and distributes a load required for call processing or communication processing to the processors A to D. The processors A to D all perform call processing with the BTSs 198a to 198b. All of the loads generated in the RNC and the loads for call processing from an external device such as the exchange 102 are distributed and processed by these processors A to D. Therefore, each of the processors A to D processes not only the load on the specific BTS 198a but also the load on the other BTS 198b. That is, all the loads are distributed and assigned so that the loads of the processors A to D are averaged.
[0009]
Further, the shared memory 203 secures resources for each of the BTSs 198a to 198b, and generates a queue for holding call information such as a mobile phone address or a telephone number of a communication partner (opposite communication device) in these resources. I have. The call information is assigned an ID (Identification: identifier), and can communicate with each other via a wired line.
[0010]
The inter-processor communication device 204 is for the processors A to D to transmit and receive signals to and from each other.
(P6) Explanation of resource acquisition by software
FIG. 20 is a diagram showing a sequence for explaining resource acquisition from the shared memory 203. The exchange 102 transmits a call control signal 1 for the terminal 1 to the control unit 201 (step H1). When the control unit 201 receives this call control signal 1, the load among the processors A to D is received. For example, the processor A is selected and the processor A is interrupted (step H2), and this call control signal 1 is notified. Upon receiving this notification, the processor A acquires a resource in the shared memory 203 (see FIG. 19) (step H3), and upon acquiring the resource, transmits the fact to the exchange 102 (call control signal 1 '[ Terminal 1])).
[0011]
On the other hand, the exchange 102 transmits the call control signal 2 for the terminal 1 after transmitting the above-mentioned call control signal 1 (step H4), and when the control unit 201 receives the call control signal 2, The task for No. 2 is distributed to the processor B (step H5) and notified to the processor B (step H6). Here, when the processor B can acquire the resource from the shared memory 203, the processor B transmits an acknowledgment (acknowledgement: response signal) to the exchange 102 as the call control signal 2 '(step H7). Thereby, the loads of the processors A to D are distributed.
[0012]
(P7) Queue management function
The queue management function is realized by a software unit (hereinafter, referred to as software) of each of the processors A to D. When the control unit 201 assigns the queue management function to the processor A, the processor A performs an exclusion process (lock) and acquires the resources of the shared memory 203 that have been assigned to the queue. During the exclusion process, the other processors BD wait.
[0013]
As an example of queue management, when the processor A acquires a queue, the processor A writes size information of the queue (queue range and queue position) to a specific area of the shared memory 203. Then, when the processing by the processor A is completed and another processor B needs to access the shared memory 203, the processor B refers to the size information written by the processor A.
[0014]
Note that a memory control state monitoring device that constantly monitors the operation state / use state of the shared buffer area is disclosed in, for example, Patent Document 2.
(P8) An example of the shared memory 203
FIG. 21B is a diagram showing an example of allocation of the shared memory 203 after a failure has occurred. The information held in the queue management information table 205 includes a queue management number (total number of queues), a next queue address, a previous queue address, and the like.
[0015]
For example, in the shared memory 203 shown in FIG. 21A, since three queues 91a to 91c are connected, the queue management number is held as 3, and the next queue address is the first queue address of the queue 91a ( The head queue address queued at the head of the queue management information table 205 is held. Note that there is no queue in front of the queue 91a even though the queue management information table 205 exists, so there is no previous queue address value.
[0016]
Each of the processors A to D can access the information such as the queue management number written in the queue management information table 205 and can easily obtain the information such as the size of the resource occupying each call. The master processor that receives a specific signal from a processor (not shown) of another device transmits a voice packet received via the control unit 201 to a processor having a small load among processors A to D. I try to sort them. In other words, the queue management information table 205 functions as a “queue terminal”.
[0017]
(P9) Explanation of operation of queue removal and queue insertion
FIGS. 23A to 23F are diagrams for explaining queue removal or queue insertion. The three-stage areas shown in these figures are provided in the queue management information table 205. In the first to third rows, the number of queue management, the first queue address, and the last queue address in the information area required for queue management are written. FIGS. 23 (a) to 23 (f) also accept this policy.
[0018]
FIG. 23A shows a main part of the queue management information table 205 immediately after initialization. In the entire area, 0 (no record) is recorded. Then, when one queue (queue of number 1) is inserted, the management number, head queue address, and last queue address of each area shown in FIG. 23B are 1, 1, and 1, respectively.
FIG. 23C shows each area when one more queue (the queue of number 2) is inserted in the state shown in FIG. 23B. That is, the total number of queues is two, and the management number, the first queue address, and the last queue address are 2 (queues with numbers 1 and 2) and 1, 2, respectively.
[0019]
FIG. 23D is a diagram showing each area when one more queue (queue of number 3) is inserted in the state shown in FIG. 23C. Therefore, the management number, the first queue address, and the last queue address are 3 (queues of numbers 1 to 3) and 1, 3, respectively.
Thus, every time a queue is inserted, its history is saved.
[0020]
Next, the main part of the queue management information table 205 at the time of removing a queue or deleting a queue will be described with reference to FIGS. 23 (e) and 23 (f). The queue deletion is recorded in the queue management information table 205 when the telephone call ends, for example, and the call is held.
FIG. 23E shows a state after the queue number 2 is extracted from the state shown in FIG. 23D. The management number is 2, the first queue address is 1, and the last queue address is the queue 3 Remains.
[0021]
FIG. 23F shows a case where the queue 2 is inserted again from the state shown in FIG. As a result, the management number is 3, the first queue address is 1, and the last queue address is 2.
(P10) Example of queue operation
In order to operate the queue, the processors A to D, when releasing the queue, for example, each of the queue management number, the next queue address, the previous queue address, and the first queue address in the queue management information table 205 shown in FIG. Need to be updated.
[0022]
On the other hand, when a failure occurs in the processors A to D for operating the queue, the recovery process is not easy. For example, when a failure occurs in the processor A performing the queue operation and a difference between the actual queue management number and the queue management number written in the queue occurs, as shown in FIG. If a difference occurs between the address and the previous next queue address, the program may not be executed normally. Therefore, in order to operate the program normally, the other processors B to D that have detected the failure perform a recovery processing operation.
[0023]
FIG. 22 is a diagram showing an example of the recovery processing operation of the shared memory 203. In the recovery processing operation shown in FIG. 22, the processors A to D search the address of the shared memory 203, trace the queue from the bidirectional queue of the head queue 91a and the final queue (tail queue) 91b, and Count and compare. If the numbers of these queues do not match, the queue address after writing to the queue management information table 205 and the previous queue address are not continuous, so that a queue is lost or the queue management information table 205 is destroyed. Is assumed. In this case, each of the processors A to D performs a recovery process to recover the queue.
[0024]
As described above, the queue management and the recovery processing are performed by the load sharing type multiprocessors A to D provided in the RNC and the shared memory 203.
By the way, when a large number of calls are input to the RNC, resources for holding queue management information are saturated, and the function of the system may be stopped. For this reason, the RNC needs to monitor the state of the memory area of the shared memory 203, and this state monitoring is realized by a state monitoring system. Hereinafter, the state monitoring system will be described.
[0025]
This status monitoring system is a memory monitoring system that monitors the resource allocation status of the shared memory 203, and monitors or manages the resource status of devices, devices, or terminals in the mobile communication system using a status monitoring table.
This status monitoring system includes a management and a recovery unit. Here, in the management, the state monitoring table divides the memory resource into pages of a predetermined size, manages whether or not each page has been updated, retains the information in a management data file, and updates the data of the updated page. In the update data file. Further, the recovery unit recovers the table data in the memory based on the information of the management data file and the update data file at the time of failure recovery.
[0026]
According to this status monitoring system, a device-to-device or device-to-terminal status (that is, a local fault occurrence status) corresponding to a portion of the status monitoring table where a fault has occurred, particularly regarding resources or buffers. Can be grasped.
[0027]
[Patent Document 1]
JP 2001-167075 A
[Patent Document 2]
JP-A-9-91172
[0028]
[Problems to be solved by the invention]
However, the system using the queue management information table 205 or the state monitoring table has the following problems (Q1) to (Q4).
(Q1) Recovery error
If a failure occurs during dequeuing (removing a specific queue), enqueuing (inserting a queue), or updating the status of the resources in the shared memory 203, the failed queue management information table 205 is destroyed and the system This can lead to processing errors. For example, when a control signal is directly input from the exchange 102 via the control unit 201, the response cannot be made, and the host device disconnects the call.
[0029]
Also, in the method of management using the status monitoring table, if a failure occurs at the time of accessing the table, data stored in the status monitoring table may be destroyed, and a system processing error may be caused.
(Q2) Processing performance due to call loss during recovery
When the signal amount (the call amount, etc.) is large, the time for the processors A to D to access the shared memory 203 becomes longer, which may cause a bottleneck. In this case, in order to reduce the processing time, the load for call processing becomes large. Also, it is more efficient to operate the recovery process only when necessary than to always perform the recovery process.
[0030]
In the above-described method of comparing the number of queues, since the processor traverses the queues in both directions, a long time for search (recovery time) is required, and another processor accesses the shared memory 203. Unavailable time increases, thus affecting the processing performance due to call loss.
(Q3) System reliability
Since the recovery process after the occurrence of a failure has an effect on a call loss or the like, the design must be designed to be fast and absolutely reliable without causing abnormal operation of the program. This increases the complexity of the system.
[0031]
Further, when recovering the destruction of the queue management information table 205, the recovery process also induces a double failure (failure of exclusive resource acquisition) by another processor and lacks the reliability of the system. In particular, the reliability of mobile communication systems in recent years is high, so reliability is an important factor.
As a method for improving system stability, a redundant configuration in which an active system and a standby system are provided is well known. For example, each of the RNCs 84a to 84b includes several active CP boards and one standby CP board. If the active CP board cannot be used for failure, maintenance, or inspection, the active NC board is used. A physical communication path between the CP board and the external device is disconnected from the active CP board and connected to the standby CP board. However, there is a problem that switching using the active system and the standby system requires time.
[0032]
In addition, the system must guarantee reliable billing and operation of devices or terminals downstream of the RNC. Therefore, when a failure occurs while the processor is accessing the shared memory 203, it is necessary to release the occupation of the shared memory 203 so that another processor (other user) can access it.
(Q4) Simplification
An increase in the system scale causes an increase in the amount of programs. The reliability and flexibility of the system each depend on the size of the system to be developed. When the conventional technique is used, processing is complicated, and reliability (program amount) and flexibility (necessary storage capacity) are bottlenecks.
[0033]
The present invention has been conceived in view of such a problem, and avoids a processing error of a system, eliminates processing performance due to a call loss at the time of recovery, and sets another processor when a failure occurs during access to a shared memory by a processor. And a queue retrieval method in a multiprocessor system, a monitoring system, and a multiprocessor system, which has simplification of processing, reliability using a large program capacity, and flexibility by allocating a relatively large working memory capacity. An object of the present invention is to provide a table recovery processing method in a multiprocessor system.
[0034]
[Means for Solving the Problems]
For this reason, the multiprocessor system of the present invention is a multiprocessor system having a plurality of processors for processing communication calls and a shared memory accessible by the plurality of processors, wherein the shared memory relates to communication calls obtained by each processor. A queue management information table for holding queue management information including a plurality of information elements for specifying the size and the position of the resource management queue for holding call information, and a plurality of update information to be written in the queue management information of the queue management information table An update history recording area for recording an update history composed of elements is provided (claim 1).
[0035]
In addition, when a failure occurs when any of the plurality of processors accesses the queue management information, the queue management information held in the queue management information table and the update recorded in the update history recording area are updated. A determination unit that determines whether recovery processing is necessary based on the history information; and a recovery unit that recovers queue management information based on the update history information when the determination unit determines that recovery is necessary. (Claim 2).
[0036]
Further, the monitoring system of the present invention is a monitoring system for monitoring a communication state between a communication device and an opposing communication device communicating with the communication device, wherein the monitoring system includes a resource status of a shared memory accessible by a plurality of processors. A state monitoring table for holding the communication state, a communication state recording area for recording the communication state held in the state monitoring table, and a local state table in the state monitoring table based on the communication state held in the communication state recording area. A recovery unit that recovers the failure occurrence part; and a selection that implements at least one of an initialization unit that initializes a local failure occurrence part in the communication state of the state monitoring table. (Claim 3).
[0037]
The queue retrieval method in the multiprocessor system of the present invention is a queue retrieval method in a multiprocessor system having a plurality of processors that process communication calls and a shared memory that can be accessed by the plurality of processors. A reference step of referring to a queue management information table that holds queue management information including a plurality of information elements that specify the size of the acquired resource management queue and the position of the queue; A recording step of recording an update history composed of a plurality of update information elements to be written to the queue management information of the queue management information table in an update history recording area, and a writing step of each processor writing the update history to the queue management information table. It is characterized by (Claim 4).
[0038]
Further, the table recovery processing method in the multiprocessor system of the present invention is a table recovery processing method in a multiprocessor system having a plurality of processors for processing communication calls and a shared memory accessible by the plurality of processors. Queue management information of a queue management information table holding queue management information including a plurality of information elements for specifying the size and the position of the resource management queue acquired by one of the processors, and the queue management information table A determination step of determining whether recovery processing is necessary based on update history information in an update history recording area that records an update history composed of a plurality of update information elements to be written in the queue management information of the queue, and determining that recovery is necessary in the determination step Is updated, based on the update history information. It is characterized in that a recovery step of recovering the queue management information (claim 5).
[0039]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings.
(A) Description of the first embodiment of the present invention
FIG. 1 is a configuration diagram of the mobile communication system according to the first embodiment of the present invention. The mobile communication system 100 shown in FIG. 1 is a system using a CDMA system, and includes low-capacity data such as telephone, text data and voice, and large-capacity data such as still images and moving images (hereinafter referred to as low-capacity data). Data and large-capacity data are referred to as information data) and transmit and receive at high speed.
[0040]
Hereinafter, each device of the mobile communication system 100 will be described with reference to FIGS. 1 to 3, and the load distribution device will be described with reference to FIG. 4.
(1) Mobile communication system 100
The mobile communication system 100 shown in FIG. 1 includes various mobile stations (MS: Mobile Station; hereinafter, referred to as MS) 10a, 10b, and 10c used by a user and, for example, 48 BTSs (radio base stations) 83a to 83c. , RNCs (Radio Network Controllers) 84a to 84b, a core network (CN: Core Network) 101, a public network (PSTN: Public Switched Telephone Networks) 85a, the Internet 85b, and a communication partner terminal (partner terminal) 11a. , 11b.
[0041]
(1-1) Core Network 101, Public Network 85a, and Internet 85b
The core network 101 includes a circuit switch and a packet switch, which are connected to each other to exchange (switch) information data. The core network 101 includes an HLR (Home Location Register) 101a that holds location registration data of the MSs 10a to 10c, and a packet switch. And a switching unit 102 comprising a Mobile Multimedia Switching System (MMS) 101b and a Gateway Mobile Multimedia Switching System (GMMS) 101c having a gateway function. The public network 85a is a subscriber line network for transmitting voice data, facsimile data and the like, and the Internet 85b is a network for transferring packets.
[0042]
Thereby, for example, the radio signal transmitted by the MS 10a is demodulated by the BTS 83a and then terminated by the RNC 84a, and the information data contained in the radio signal is transmitted to the public network 85a or the Internet via the core network 101. 85b. The information data transmitted by the partner terminal 11a is transferred to the RNC 84a based on the location registration data of the HLR 101a via the public network 85a, the Internet 85b and the core network 101, and is modulated by the BTS 83a into a radio signal. Sent to MS 10a.
[0043]
Note that the mobile communication system 100 can also use the cdma2000 system or the W-CDMA system.
(1-2) RNC (Base Station Controller) 84a-84b
The RNCs 84a to 84b control a plurality of BTSs 83a to 83c, multiplex and transmit and receive a plurality of channel data between each of the BTSs 83a to 83c and the core network 101, and perform high-speed and large-capacity data processing. Requires ability.
[0044]
Each of the RNCs 84a to 84b has a function of transferring information data between the BTSs 83a to 83c and the core network 101, and a function of transmitting and receiving a control signal for call connection processing or maintenance operation for each of the BTSs 83a to 83c. And has functions such as channel assignment, handover execution, and incoming / outgoing connection. Thereby, the information data from the public network 85a or the Internet 85b is transferred to any of the plurality of BTSs 83a to 83c according to the destination, and the information data from the plurality of BTSs 83a to 83c is transmitted to the public network 85a or 83b. The data is transferred to the Internet 85b.
[0045]
Each of the RNCs 84a to 84b, the plurality of BTSs 83a to 83c, and the plurality of MSs 10a to 10c transmit and receive control signals and data signals by using a wireless protocol, and function as a radio access network (RAN: Radio Access Network). ing. On the other hand, the RNCs 84a to 84b and the core network 101 transmit and receive information data by wire.
[0046]
(1-3) Multiprocessor system of the present invention
This multiprocessor system is applied to a load distribution device provided in each of the RNCs 84a to 84b. The information data transferred by the RNCs 84a to 84b has a high speed and a large capacity and processes information data from both the plurality of BTSs 83a to 83c and the core network 101. Therefore, the load on each of the RNCs 84a to 84b is extremely large. large. For this reason, a load distribution function is provided in each of the RNCs 84a to 84b so that all loads on information data are distributed.
[0047]
It should be noted that the multiprocessor system of the present invention can also be applied to units of devices such as calls, MSs 10a to 10c, network terminals, computers, and BTSs 83a to 83c.
(1-4) MS10a etc.
The MSs 10a to 10c wirelessly communicate information data and control data related to communication and control with the BTSs 83a to 83c, and include a mobile phone 10a, a video playback device 10b, and a videophone 10c.
[0048]
For example, as shown in FIG. 2, the mobile phone 10a outputs and receives information data obtained by encoding signals from the voice receiving unit (SPK) 86a, the voice transmitting unit (MIC) 86g, and the voice transmitting unit 86g. A baseband processor 86b for inputting decoded data to a voice receiver 86a or a display (not shown); and a modulation / demodulation for modulating information data from the baseband processor 86b into a radio signal and demodulating a received radio signal. (TRX) 86c, a transmission / reception amplifier (T / R AMP) 86d for amplifying a radio signal from the modem unit 86c and amplifying a reception signal, and a key information holding unit 86e for holding terminal identification information. And a mobile unit control unit (control unit) 86f for controlling the MSs 10a to 10c.
[0049]
As for the video reproducing device 10b and the videophone 10c, their wireless transmission / reception units are equivalent to the modulation / demodulation unit 86c of the mobile phone 10a and the like, and the description thereof will be omitted. In the following, a description will be given using an MS (mobile phone) 10a.
(1-5) Counterpart terminals 11a and 11b (see FIG. 1)
The partner terminals 11a and 11b both communicate with the MS 10a or transmit and receive information data, and are, for example, fixed telephones, mobile telephones, or personal computers (PCs).
[0050]
(1-6) BTS83a-83c
FIG. 2 is a functional block diagram of the mobile communication system 100 according to the first embodiment of the present invention. In FIG. 2, components having the same reference numerals as those described above have the same functions.
Each of the 48 BTSs 83a to 83c transmits and receives wireless data to and from the MS 10a and transmits and receives wired data to and from the RNCs 84a to 84b. The transmitting and receiving amplifier 86d amplifies wireless signals with high power and amplifies received signals with low noise. A modulation / demodulation unit 86c for modulating / demodulating a radio signal, a baseband processing unit 86b for processing a baseband signal, and a highway-in control unit (HW-CNT) 87a for controlling data transmission / reception between the RNCs 84a to 84b and the BTSs 83a to 83c. And a base station control section (control section) 87b for monitoring and controlling the inside of the BTSs 83a to 83c.
[0051]
Further, the base station control unit 87b may be configured by providing a shared memory equivalent to the shared memory 3 included in the RNC 84a (not shown).
Each of the BTSs 83b and 83c has the same configuration as the BTS 83a, and redundant description will be omitted. The BTS number 48 is an example, and is not limited to this value.
[0052]
(2) Configuration of RNCs 84a-84b
The RNCs 84a to 84b include a load distribution device (CONT) 1 having a control function and a load distribution function for call processing (Call Processing), a line interface (LIF: Line Interface) 85a for terminating a line with the BTSs 83a to 83c. , BTS 83a to 83c and an ATM switch (Asynchronous Transfer Mode Switch) 85b for switching a packet from the exchange 102 based on its destination or a predetermined path, an output interface 85c for terminating a line to the exchange 102, and call processing. And the like, and a terminating unit (SU) 85d for terminating a control signal such as the above.
[0053]
The terminating unit 85d has a mobile station facing signal terminating unit (MSU), an external device facing signal terminating unit (ESU), and an OPS facing signal terminating unit (OSU), and receives signals from the MS 10a and the exchange 102. An input is made to the load distribution device 1.
(3) Load distribution device (multiprocessor system) 1
FIG. 3 is a schematic block diagram of the RNCs 84a to 84b according to the first embodiment of the present invention. Those having the same reference numerals as those described above in FIG. 3 have the same functions.
[0054]
(3-1) Load distribution device 1
The functions of the load balancer 1 are as follows: communication between external devices such as the BTSs 83a to 83c and the exchange 102 (see FIG. 1) and each module of the RNCs 84a to 84b itself, transmission and reception of control signals for call connection, maintenance operation, and transmission and reception of these signals. This is load distribution for processing. The load distribution device 1 includes an inter-processor communication device (bus control unit: BCONT [Bus Controller]) 5, call processing units (CP: Call Processing) 2a to 2e, a shared memory (CM: Common Memory) 3, It comprises a hard disk drive (HDD: Hard Disk Drive) 4, a debugger (DB: Debugger) 5f, and a communication bus 5e.
[0055]
(3-2) Call processing units 2a to 2e
The functions of each of the call processing units 2a to 2e are assignment of a wireless channel, call information processing for an incoming call, conversion processing between a global address and a local address, and main control. A processor 22a, a ROM (Read Only Memory) 23, and a RAM (Random Access Memory) 24 are provided.
[0056]
FIG. 4 is a diagram illustrating a main part of the load distribution device 1 according to the first embodiment of the present invention. The load distribution apparatus 1 shown in FIG. 4 includes a plurality of processors 22a to 22e that process communication calls and a load distribution control that distributes communication calls from an external device such as the exchange 102 according to the loads on the processors 22a to 22e. (Distribution control unit) 9 and a shared memory 3 accessible by the processors 22a to 22e. The shared memory 3 has a queue management information table (see FIG. 5 and the like) 7 for holding queue management information on a resource management queue for holding call information on communication calls, and an update history of queue management information in the queue management information table. And a memo recording unit (update history recording area, see FIG. 8) 8 for recording the information.
[0057]
Note that the function of the call processing unit 2a is exerted by software in which the processor 22a, the ROM 23, and the RAM 24 cooperate. Similarly, the respective functions of the call processing units 2b to 2e are also provided with the respective processors 22b to 22e on the respective CP boards, and the ROM 23 and the RAM 24 are exhibited in cooperation. The call processing units 2a to 2e cooperate to distribute the total load of the 48 BTSs 83a to 83c for the call processing.
[0058]
(3-3) Inter-processor communication device 5
The inter-processor communication device 5 has a control function for allowing the processors 22a to 22e to transmit and receive signals to and from each other. The load distribution device 1 is configured such that, for example, the processor 22a among the processors 22a to 22e is a master processor and receives specific signals from the other processors 22b to 22e. In other words, all of the other processors 22b to 22e use the inter-processor communication device 5 to transmit and receive signals to and from the master processor 22a.
[0059]
(3-4) Load distribution control unit (distribution control unit) 9
The distribution control unit 9 receives information data and control data from the exchange 102 and the BTSs 83a to 83c, and distributes the load of communication call processing to the call processing units 2a to 2e. The load distribution function is performed by, for example, the master processor 22a. Note that the other processors 22b to 22e can also realize the load distribution function.
[0060]
(3-5) Shared memory 3
The information held in the shared memory 3 is information that the plurality of processors 22a to 22e need to take over each other or information that must not be lost. For example, if a failure occurs when the MS 10a and the partner terminal 11a enter a call state such as "Hello" or "Yes Yes", the call state between the two is maintained because the call state between the two is maintained. Must be taken over by the other processors 22a to 22e.
[0061]
Each processor processes a call by assigning a task according to the load in order to distribute the load.
FIG. 5 is a diagram schematically showing an area of the shared memory 3 according to the first embodiment of the present invention. The shared memory 3 shown in FIG. 5 is shared and accessed by the processors 22a to 22e, and holds a queue management information table 7, resources (queues) 6a to 6c, 6q, and other data. And a memo recording unit 8.
[0062]
The resources 6a to 6c are memory areas obtained to hold call information A to C relating to a plurality of calls, respectively, and a plurality of queues A to C are generated in this memory area. The resource 6q is also a memory area acquired to hold call information D relating to a plurality of calls, and a queue D is generated. Since each of the RNCs 84a to 84b receives a large number of calls, one resource is allocated to one device in order to efficiently manage the large number of calls. That is, the shared memory 3 performs resource allocation for each of the BTSs 83a to 83c by a resource allocation function generally provided.
[0063]
For example, in the shared memory 203 shown in FIG. 21A, resources A to C corresponding to three BTSs 83a to 83c are allocated, and a plurality of received calls are inserted into queues A to C, respectively. Also, it is taken out (dequeue). Accordingly, the queues A to C have different numbers of inserted queues and different ranges of queue areas, and the queue management information table 7 holds queue management information on queues that manage resources holding call information. is there.
[0064]
(3-6) Queue management information table 7
The queue management information table 7 holds queue management information including a plurality of information elements that specify the size and position of a queue that manages resources holding call information acquired by the processors 22a to 22e. This information element is, for example, a queue management number (total number of queues), a first queue address, a previous queue address and a rear queue address, and may also include the last queue address.
[0065]
Thus, queue management can be efficiently performed by using each information element. The queue management information table 7 holds queue management information including at least a queue management number and a head queue address for a queue that manages resources holding call information acquired by the processors 22a to 22e. This is because, according to these two types of elements, each queue can be specified.
[0066]
(3-7) Memo recording unit 8
Next, the memo recording unit 8 (see FIG. 8) records an update history of the queue management information in the queue management information table 7. That is, the memo recording unit 8 records a plurality of update information elements (for example, queue management number, head queue address, previous queue address, and rear queue address) to be written in the queue management information of the queue management information table 7. The memo recording unit 8 secures, for example, three memo records (memo recording units) 1, 2, and 3, and records the queue management number and the like. For example, the memo records 1 and 2 can temporarily write the queue management number and the head queue address, respectively, and the memo records 1, 2, and 3 respectively store the queue number to be taken out, the previous queue address, The queue address can also be written temporarily. That is, the memo recording unit 8 records an update information element to be written in the queue management number and the head queue address held in the queue management information table 7.
[0067]
Upon receiving the control signal from the distributed control unit 9, each of the call processing units 2a to 2e (see FIG. 4) acquires a physical resource from the shared memory 3, generates a queue for the obtained resource, and generates the queue. The information on the queue that has been set is centrally held in the queue management information table 7.
Thus, each of the processors 22a to 22e refers to the queue and performs signal processing while updating the queue. For example, when the call processing unit 2a is overloaded, the other call processing units 2b to 2e perform data processing.
[0068]
(3-8) Clock unit 88 and the like
The RNC (see FIG. 3) multiplexes a clock unit (CLK) 88 for generating a reference timing, a trunk card DHT (Diversity Handover Trunk card) 89a for performing diversity handover processing, and a MAC (Media Access Control) layer of a wireless line. It also has a Mac demultiplexing card M-MUX (Mac Multiplexer) 89b and BWC for processing separation.
[0069]
Also, the HDD 4a (see FIG. 3) holds information obtained about the firmware. The debugger 5f writes and reads data from and to the shared memory 3. By outputting a signal for writing and reading to the bus 5e, the writing and reading can be performed.
(3-9) Example of implementation
The load distribution device 1 is provided on a substrate (backboard) on a rear portion of a device rack of the RNCs 84a to 84b.
[0070]
FIG. 6 is a schematic diagram of the load distribution device according to the first embodiment of the present invention as viewed from above. In the load distribution device 1 shown in FIG. 6, the functions of the interprocessor communication device 5, the call processing units 2a to 2e, the debugger 5f, the shared memory 3, the clock unit 88, and the power supply unit 88a are provided on a planar backboard 82. The substrate is inserted vertically. That is, each function is realized for each board, whereby the burden of maintenance and the like is reduced, and it is possible to flexibly respond to changes in system specifications. The shaded portion shown in FIG. 6 represents the upper surface of the backboard 82.
[0071]
Each substrate is assigned an ID (Identification), and can communicate with each other via a signal transmission path. Thus, the processors can communicate with each other via the shared memory 3.
As an example of this mutual communication, each of the processors 22a to 22e obtains an ID included in the data output to the bus 5e, and compares the obtained ID with an ID stored in a register (not shown) or the like in advance. , It is determined whether to use the data.
[0072]
Each board is detachable, and the number of CP boards can be increased or decreased according to the processing capacity of the call processing units 2a to 2e. The load distributing apparatus 1 is provided with a total of six CP substrates, that is, five active CP substrates and one spare CP substrate, and can cope with 48 BTSs 83a to 83c. A spare CP board (not shown) is provided in case one of the five CP boards fails.
[0073]
The RNCs 84a and 84b can use a redundant configuration including a plurality of active (ACT: ACT) CP substrates and a standby (STBY) CP substrate. Here, the standby CP board is provided for the failure or maintenance of the active CP board, and if any of the active CP boards fails, the physical communication path connected to the external device is lost. Is disconnected from the failed CP substrate and connected to the standby CP substrate. As a result, a load distribution type multiprocessor system is configured.
[0074]
Further, resources corresponding to each of the 48 BTSs 83a to 83c are secured in the shared memory 3 of the CM board, and the processors 22a to 22e and the shared memory 3 cooperate, so that the CP boards of the BTSs 83a to 83c are At the time of overload, the CP boards for the other BTSs 83a to 83c perform data processing of the BTSs 83a to 83c.
[0075]
As described above, the load distribution device 1 can process a very large amount of call processing data, and can recognize and appropriately process a call to be processed from a large amount of calls, thereby achieving data multiplexing and data separation. It becomes possible.
(3-10) Recovery processing
When a failure occurs in the queue operation processors 22a to 22e, a difference between the actual queue management number and the queue management number written in the queue, or a difference between the next queue address and the previous next queue address occurs.
[0076]
Each of the processors 22a to 22e has a detection function of detecting occurrence of a failure, and in the present embodiment, the failure detection function of the master processor 22a operates. More specifically, the master processor 22a triggers the recovery process by an external interrupt from hardware. For example, using the software by the master processor 22a, the master processor 22a controls the status of both the queue management information table 7 and the memo recording unit 8. Compare.
[0077]
Then, the master processor 22a that has detected the failure performs a recovery process so that the program is executed normally. This recovery processing function is realized by, for example, the call processing unit 2a and the shared memory 3, but can also be realized by the other call processing units 2b to 2e (software by the processors 22b to 22e) and the shared memory 3.
[0078]
(3-11) Judgment unit (judgment unit) 25a and restoration unit (restoration unit) 25b
Both the determination unit 25a and the restoration unit 25b must be provided in each of the call processing units 2a to 2e. This means that even if a failure occurs when any of the processors 22a to 22e accesses the shared memory 3, the processors 22a to 22e as hardware capable of performing normal recovery processing and the call processing unit as software This is because 2a to 2e are secured.
[0079]
FIG. 7 is a schematic block diagram of the determination unit and the recovery unit according to the first embodiment of the present invention. The determination unit 25a and the recovery unit 25b shown in FIG. 7 are all call processing units 2a to 2e. It is provided in.
Here, when any one of the processors 22a to 22e encounters a failure when accessing the queue management information, the determination unit 25a determines the queue management information held in the queue management information table and the memo recording. Based on the update history information recorded in the unit 8, the necessity of the recovery process is determined.
[0080]
The call processing unit 2a and the shared memory 3 cooperate to perform the functions of the determination unit 25a and the recovery unit 25b.
The recovery unit 25b recovers the queue management information based on the update history information when the determination unit 25a determines that the recovery process is necessary.
The acquisition and release of the resources of the shared memory 3 are always queue-managed, and the processor 22a accesses information elements (queue management number, head queue address, previous queue address, and rear queue address) necessary for queue management. When a failure occurs, the determination unit 25a determines whether the queue management information table 7 of the shared memory 3 has been destroyed based on the contents of the memo recording unit 8. The determination unit 25a causes the recovery unit 25b to perform a complete recovery process when the queue management information table 7 has been destroyed, and causes the recovery unit 25b to execute the complete recovery process when the queue management information table 7 has not been destroyed. Do not perform recovery processing.
[0081]
As described above, since each of the processors 22a to 22e refers to the queue management information in the shared memory 3 based on the memo record and recovers the information, it is possible to recover the queue management information table 7 in a very short time or instantaneously.
(4) Description of operation
The recovery processing method of the queue management information table 7 of the shared memory 3 according to the present invention having the above configuration will be described in detail.
[0082]
When receiving the control signal, the load distribution device 1 distributes the load to each of the processors 22a to 22e, and manages the resources of the shared memory 3 using queues. When accessing the queue management information table 7, each of the processors 22a to 22e first temporarily writes the contents to be written to the queue management information table 7 to the memo recording unit 8 of the shared memory 3, and stores the written contents. I do.
[0083]
Here, in a case where data is held in a plurality of queues A to D, for example, when operating a queue C among the plurality of queues A to D, each processor 22a to 22e accesses the queue C. When a failure occurs in any one of the processors 22a to 22e, a recovery process for recovering the failure is performed.
[0084]
Here, each of the RNCs 84a to 84b may be provided with two load balancers 1, and in this case, after the occurrence of the failure is detected, first, the active processor and the standby processor Are mutually changed.
A failure occurs while the queue management information table 7 in the shared memory 3 is being accessed by each of the processors 22a to 22e. After recovery, the queue management number or the head queue address in the queue management information table 7 does not match the entity. May have occurred. In this case, the queue management information table 7 is recovered based on the memo information recorded at the time of access.
[0085]
Hereinafter, a method of extracting the head queue and a recovery processing method will be described. The queue retrieval method in the multiprocessor system of the present invention is for a multiprocessor system (for example, the load distribution device 1) having processors 22a to 22e that process communication calls and a shared memory 3 that can be accessed by the processors 22a to 22e. .
[0086]
First, queue management that holds queue management information including a plurality of information elements that specify the size of the resource management queue and the position of the queue acquired by any one of the processors 22a to 22e (for example, the processor 22a itself). Refer to the information table (reference step). That is, the queue management number, the head queue address, the previous queue address, and the rear queue address are referred to.
[0087]
Next, each of the processors 22a to 22e records an update history including a plurality of update information elements to be written in the queue management information of the queue management information table 7 in the memo recording unit (update history recording area) 8 (recording step). That is, the updated queue management number, head queue address, previous queue address, and rear queue address are recorded in the memo recording unit 8.
[0088]
Then, each of the processors 22a to 22e writes the update history into the queue management information table 7 (write step).
The queue removal method will be described in more detail.
For the reference, the processors 22a to 22e refer to the queue management number, the head queue address, the queue management number among the previous queue address and the rear queue address, and the head queue address as information elements.
[0089]
Next, in the recording, each of the processors 22a to 22e records the updated queue management number (first update value) or the updated head queue address (second update value) as an update information element.
Then, upon the writing, each of the processors 22a to 22e writes the updated queue management number or the updated head queue address to the queue management information table 7.
[0090]
Hereinafter, a method of removing a queue will be specifically described.
(5) A method of taking out the head queue.
A method of deleting the head queue (resource) will be described with reference to FIGS. Here, steps J1 to J5 shown in FIG. 8 correspond to steps J1 to J5 shown in FIG.
[0091]
FIG. 8 is a diagram for explaining the head queue extracting method according to the first embodiment of the present invention. In FIG. 8, a memo recording unit 8, a queue management information table 7, and cues A to D are displayed, and the upper half of the sheet is a memo recording unit 8, a queue management information before cue removal. Table 7 shows the status of the queue. The lower half represents their state after removal of the queue. Note that the direction from the top to the bottom represents the passage of time. Here, the queues A to D are generated before the queue is taken out, and the head queue A is taken out from the queues A to D. The queue management information table 7 holds the head queue address and the total number of queues forming the queue columns A to D.
[0092]
FIG. 9 is a flowchart for explaining the head queue removal processing according to the first embodiment of the present invention. The head queue removal process will be described with the processor 22a as an example. The processing for the processors 22b to 22e is the same as the processing for the processor 22a.
(J1) The processor 22a extracts the queue (resource) queued at the head. When taking out the head queue, the processor 22a refers to the queue A by referring to the head queue address from the queue management information table 7.
[0093]
(J2) The processor 22a sets the next queue address of the queue (resource) queued at the head of the memo record 2. The next queue address B of the queue (resource) queued at the head is taken out and set in the memo record 2.
(J3) The processor 22a holds the queue management number (total number) in the memo record 1, and sets 3 obtained by subtracting 1 from the queue management number 4 in the queue management information table 7 in the memo record 1.
[0094]
In step J2 or step J3, if a failure occurs while the processor 22a is rewriting the memo records 1 and 2 (step J6a), no recovery is performed (step J6b).
(J4) The processor 22a rewrites the head queue address of the queue management information table 7 from the queue A to the queue B address.
[0095]
(J5) The processor 22a subtracts 1 from the queue management number in the queue management information table 7 by taking out the queue A and sets it to 3.
In step J4 or step J5, when a failure occurs while the processor 22a is rewriting the queue management information table 7 (step J7a), the determination unit 25a determines that the recovery process is necessary, and , 2 are rewritten (step J7b).
[0096]
This simplifies the recovery process and realizes high-speed recovery.
(6) Recovery processing method
The table recovery processing method in the multiprocessor system of the present invention is for a multiprocessor system having a plurality of processors 22a to 22e for processing communication calls and a shared memory 3 accessible by the processors 22a to 22e.
[0097]
First, for example, the processor 22a among the processors 22a to 22e specifies the size of the resource management queue acquired by itself and the position of the queue (queue management number, head queue address, previous queue address, and subsequent queue address). Queue management information of the queue management information table 7 holding queue management information composed of information elements, and update information elements (queue management number, head queue address, etc.) such as the updated queue management number to be written in the queue management information of the queue management information table 7 , A previous queue address and a rear queue address), and determines whether or not a recovery process is necessary, based on the update history information of the memo recording unit 8 that records an update history including the update history (determination step).
[0098]
Then, when any of the processors determines that the recovery is necessary in the determination, the processor recovers the queue management information based on the update history information (recovery step).
As a result, the plurality of processors 22a to 22e cooperate, so that the system is robust against failure.
[0099]
Next, the table recovery processing method will be described in more detail.
In determining whether or not the recovery processing is necessary, any one of the plurality of processors 22 a to 22 e determines the queue management number or the head queue address of the queue management information held in the queue management information table 7, and the memo recording unit 8. The necessity is determined based on the queue management number or the match / mismatch with the head queue address.
[0100]
Upon recovery, one of the processors rewrites the queue management number or head queue address of the memo recording unit 8 to the queue management number or head queue address of the queue management information table 7 based on the necessity.
In this way, recovery processing after a failure can be performed quickly.
Next, a recovery processing method when a failure occurs during the removal of the head queue will be described with reference to FIG.
[0101]
FIG. 10 is a flow chart for explaining the recovery processing for removing the head queue according to the first embodiment of the present invention. Each of the processors 22a to 22e can perform the recovery process described below, and the processor 22a will be described, and the detailed description of the processors 22b to 22e will be omitted. Here, reference numerals 1 to 6 shown in FIG. 10 correspond to the following (6-1) to (6-6), respectively.
[0102]
(6-1) The determining unit 25a determines whether the head queue address of the memo record 2 matches the head queue address of the queue management information table 7.
(6-2) If the judgment results do not match, the judgment unit 25a goes through the “mismatch” route, rewrites the memo record 2, and before the information in the queue management information table 7 is rewritten, a failure has occurred. It has not been done. Therefore, the recovery process is not performed ignoring the memo record.
[0103]
(6-3) The judgment unit 25a indicates that the memo record 1 is rewritten when the judgment results match, and that the memo record 1 is rewritten when the information in the queue management information table 7 is normal when the judgment results match. I have.
(6-4) The determination unit 25a determines whether the queue management number of the memo record 1 matches the queue management number of the queue management information table 7.
[0104]
(6-5) When the determination results do not match, the determination unit 25a rewrites the memo record 1 and, when a part of the queue management information table 7 is normal, has not been rewritten. Therefore, recovery is performed by referring to the memo record 1 and rewriting the queue management number in the queue management information table 7 to 3.
(6-6) When the determination results match, the determination unit 25a is in a state where the extraction has been completed when the queue management information table 7 and the head queue are normal. Therefore, recovery becomes unnecessary.
[0105]
(6-7) If the determination result of the determination unit 25a is “mismatch” in the above (6-1), a failure occurs before the queue management information table 7 is written through the route marked “mismatch”. Therefore, the determination unit 25a determines that the data is not damaged, and does not perform the recovery process.
(6-8) When the determination unit 25a determines that the queue management numbers do not match in (6-4), the head queue address of the queue management information table 7 has been rewritten through the NO route, It is determined that a failure has occurred.
[0106]
(6-9) The determination unit 25a performs a recovery process of replacing the queue management number of the memo record with the queue management number in the management number queue management information table 7.
Note that the head queue removal operation of the processors 22b to 22e is the same as the operation of the processor 22a.
As described above, according to the present recovery processing method, the acquisition and release of the resources of the shared memory 3 are queue-managed, and the information areas such as the queue management number required for this queue management, the first queue address, the previous queue address, and the rear queue address. Is stored as the memo recording unit 8 before it is accessed. Thus, the recovery process can be performed in a single time and at a high speed. Thus, the induction of double failure is avoided.
[0107]
Then, with regard to the queue management information table 7, it is possible to perform processing as if it were just to record a memo, and it is possible to perform recovery using the minimum data required for recovery processing, thereby simplifying the system.
(7) A method for extracting a queue located between a plurality of queues
With reference to FIG. 11 and FIG. 12, a method of taking out and deleting a queue C queued between the queues A to D (hereinafter, referred to as an intermediate queue C in FIGS. 11 and 12) will be described. Note that the processor 22a will be described, and redundant description of the processors 22b to 22e having the same functions as the processor 22a will be omitted. The symbols K1 to K6 shown in FIGS. 11 and 12 correspond to each other.
[0108]
FIG. 11 is a diagram for explaining a method of extracting the intermediate queue C according to the first embodiment of the present invention. The queue management information table 7 holds the address of the first queue A.
FIG. 12 is a flowchart for explaining the process of taking out the intermediate queue C according to the first embodiment of the present invention.
[0109]
(K1) The processor 22a extracts the queue C on the way.
(K2) The processor 22a acquires the address of the first queue A in the queue management information table 7, sequentially follows the queue from the first queue A to the queue D, refers to the intermediate queue C, and records the memo. The intermediate queue C is set to 1.
(K3) The processor 22a writes the address of the previous queue B of the intermediate queue C to the memo recording 2, and takes out the intermediate queue C. For this reason, the queue next to the queue B is changed from the intermediate queue C to the queue D. Therefore, the processor 22a sets the address of the previous queue B of the intermediate queue C to the address (B) in the memo record 3.
[0110]
(K4) The queue D is changed by removing the halfway queue C, so the processor 22a sets the next queue address (D) of the halfway queue C in the memo recording unit 3.
In each of steps K2, K3, and K4, if a failure occurs while the processor 22a is rewriting the queue management information table 7 (step K7a), the determination unit 25a determines that the recovery process is unnecessary (step K7b). ).
[0111]
(K5) The processor 22a rewrites the address of the next queue D of the queue B from C to D in the memo recording unit 3.
(K6) The processor 22a rewrites the previous queue address of the queue D from C to B.
In step K5 or step K6, if a failure occurs while the processor 22a is rewriting the queue management information table 7 (step K8a), the determination unit 25a determines that the recovery process is necessary, and the recovery unit 25b Rewrites the contents of the memo records 1 to 3 into the queue management information table 7 (step K8b).
[0112]
The operation of the processors 22b to 22e to take out the intermediate queue C is the same as the operation of the processor 22a.
(8) Recovery processing method
With reference to FIG. 13, a description will be given of a recovery processing method when a failure occurs during the removal of the intermediate queue C.
[0113]
FIG. 13 is a flowchart for explaining a recovery process for removing a halfway queue according to the first embodiment of the present invention. The operation of the determination unit 25a described below is the same as that of the processor 22a. Reference numerals 1 to 7 shown in FIG. 13 correspond to (8-1) to (8-7) shown below, respectively.
(8-1) The determination unit 25a derives the middle queue C from the information of the memo record 1, determines the match / "mismatch" between the previous queue address of the middle queue C and the previous queue address of the memo record 2, and The determination unit 25 a determines whether the next queue address matches the next queue address of the memo recording unit 3.
[0114]
(8-2) When the judgment results do not match
Since the queue has already been dequeued, the previous queue address of the memo recording 2 does not match the previous queue address of the intermediate queue C, and the next address of the memo recording unit 3 does not match the next queue address of the intermediate queue C. In this case, no recovery is performed through the NO route.
[0115]
(8-3) When the judgment results match
The determination unit 25a determines whether the address of the queue C in the middle of the memo recording 1 and the next queue address of the queue B are the same through the YES route.
(8-4) When the judgment results do not match
The process ends through the NO route without performing recovery.
[0116]
(8-5) When the judgment results match
In the determination in (8-3), the determination unit 25a determines whether the midway queue C address of the memo recording 1 is the same as the previous queue address of the queue D.
(8-6) When the judgment results do not match
This is the case where the determination unit 25a has not completed the removal, and the intermediate queue C is set to the next queue address of the queue B via the NO route. As a result, the recovery process is performed.
[0117]
(8-7) When the judgment results match
The removal process is not executed, and the dequeue process is not performed through the YES route, so that the recovery process is unnecessary.
(9) Comparison between the conventional table recovery processing method and the present table recovery processing method.
[0118]
With respect to the queue management information table 7 at the time of occurrence of a failure, the conventional table recovery processing method is such that each of the processors 22a to 22e performs bidirectional communication between an intermediate queue connected to the first queue and an intermediate queue connected to the last queue. , A plurality of connected queues are traced, the number of bidirectional queues is counted, and the queue management information table 7 is recovered based on the success or failure of the count value. Therefore, the processors 22a to 22e required much time to recognize the occurrence of the failure.
[0119]
On the other hand, according to the present table recovery processing method, each of the processors 22a to 22e manages the queue of acquiring and releasing the resources of the shared memory 3, and the processors 22a to 22e perform the processing necessary for the queue management. If a failure occurs when accessing the information area, the recovery unit refers to the queue management information in the shared memory 3 based on the memo record and recovers the information. Can recover.
[0120]
(10) Queue insertion method
A method of inserting the queue D will be described with reference to FIGS.
FIG. 14 is a diagram for explaining a method of inserting a queue according to the first embodiment of the present invention, and shows a method of inserting and connecting a queue D to three queues A to C. The queue management information table 7 holds the head queue address and the total number of queues connected to the queue.
[0121]
The enqueue method will be described, for example, in the case where the processor 22a uses it. However, the head queue fetching operation of the processors 22b to 22e is the same as the operation of the processor 22a.
FIG. 15 is a flowchart illustrating the enqueue process according to the first embodiment of the present invention.
[0122]
(10-1) Creation of queue D
The processor 22a holds the last (END) as the next queue address of the queue D because it becomes the last queue (step L1).
(10-2) The processor 22a holds the address of the queue C in the memo record 2 (step L2).
[0123]
(10-3) The processor 22a sets the address of the queue D to be enqueued in the memo recording unit 3 (Step L3).
(10-4) The processor 22a holds the queue management number (total number) in the memo record 1. Here, by connecting the queue to the memo record 1 from the queue management number in the queue management information table 7, 1 is added (+1), and 4 is set (step L4).
[0124]
(10-5) The processor 22a rewrites the next queue address of the queue C
The queue C to be changed is searched by tracing the queue from the head queue address in the queue management information table 7, and the next queue address is rewritten from the end (END) to D (step L5).
(10-6) Rewrite the queue management number in the queue management information table 7 The processor 22a sets the queue management number in the queue management information table 7 to 4 by connecting the queue D (step L6).
[0125]
In steps L2 to L4, if a failure occurs while the processor 22a is rewriting the memo records 1 and 2 (step L6a), the recovery is not performed (step L6b).
Further, in step L5 or step L6, when a failure occurs while the processor 22a is rewriting the queue management information table 7 (step L7a), the determination unit 25a determines that the recovery process is necessary, and , 2 are rewritten (step L7b).
[0126]
(11) Enqueue recovery processing method
FIG. 16 is a flowchart for explaining an enqueue recovery process according to the first embodiment of the present invention. A recovery processing method when a failure occurs during enqueue will be described. Reference numerals 1 to 6 shown in FIG. 16 correspond to (11-1) to (11-6) shown below, respectively.
[0127]
(11-1) The determination unit 25a determines whether the next queue address of the queue C of the memo recording unit 2 matches the next queue address of the queue C of the memo recording unit 3.
(11-2) When the determination results do not match
The determination unit 25a determines that the queue C is not damaged through the NO route, and does not perform the recovery process. That is, since a failure has occurred before writing the next queue address of the queue C, the queue is not damaged, so that it is determined that the recovery process is unnecessary, and the process ends.
[0128]
(11-3) When the judgment results match
In this case, the determination unit 25a passes the YES route, and when the next queue address information of the memo recording unit 3 and the queue C is normal, it corresponds to a rewritten state.
(11-4) The determining unit 25a determines whether the queue management number of the memo record 1 matches the queue management number of the queue management information table 7.
[0129]
(11-5) When the judgment results do not match
In this case, the determination unit 25a corresponds to a state in which a failure has occurred after the next queue address of the queue C has been rewritten, and the queue management information table 7 has not been rewritten. Therefore, the determination unit 25a performs a recovery process of rewriting the queue management number in the queue management information table 7 to 4 by referring to the memo record 1 through the NO route.
[0130]
(11-6) When the judgment results match
In this case, since the rewriting of the queue management information table 7 and the queue C has been completed normally, the determination unit 25a passes the YES route and ends the processing without the need for the recovery processing.
Thus, the enqueue recovery process is performed.
[0131]
In addition, it is possible to realize a highly reliable system that operates at high speed because of the simplified processing method.
(B) Description of the second embodiment of the present invention
In the shared memory 3, when the queue indicating the call that has ended the call is deleted, the vacant resources (memory area or buffer area) are released, and the released resources are used as the call processing units 2a to 2e ( Alternatively, it is obtained by the processors 22a to 22e). For this reason, the size of the resources of the shared memory 3 is not fixed for each of the call processing units 2a to 2e and varies.
[0132]
Here, if a failure occurs when the call processing units 2a to 2e delete a queue that is no longer needed due to the termination of the call, the queue management information table 7 is destroyed. This causes a situation in which the section remains occupied, and also causes a decrease in resources allocated to each processor 22a to 22e.
For this reason, the mobile communication system according to the second embodiment uses the status monitoring table to determine the resource status or communication status of each communication device (opposite communication device) such as the MS 10a, the BTSs 83a to 83c or the RNCs 84a to 84b in the mobile communication system. Distributed monitoring or distributed management.
[0133]
FIG. 17 is a configuration diagram of a mobile communication system according to the second embodiment of the present invention. The mobile communication system 100a shown in FIG. 17 uses a CDMA system, and includes an RNC (communication device) 84p, a plurality of BTSs (opposite communication devices) 83p communicating with the RNC 84p, and an MS (mobile device). 10p.
Here, the MS 10p is provided with a control unit 86p having a shared memory 29 that can be accessed by a plurality of processors 22a to 22e, and the BTS 83p is provided with a mobile device control unit 87p having a shared memory 29. Further, the RNC 84p is provided with a distributed control unit 1a having a shared memory 29. Each of the MS 10p, the BTS 83p and the RNC 84p allocates a state monitoring table to the shared memory 29. In FIG. 17, those having the same reference numerals as those described above represent the same ones.
[0134]
The status monitoring table holds the above communication status including the resource status of the shared memory 3 that can be accessed by the plurality of processors 22a to 22e. This state monitoring table stores, in the shared memory 29, a queue address and a use state of a storage area corresponding to the queue address in association with each other.
FIG. 18 is a diagram schematically showing an area of the shared memory 29 according to the second embodiment of the present invention. The shared memory 29 shown in FIG. 18 includes a state monitoring table 30 for holding a communication state including a resource state, communication state holding units 31a to 31c for holding communication states of a plurality of BTSs 83a to 83c, and a state monitoring table 30. And a memo recording unit (communication state recording area or memo recording area) 32 for recording the communication state held in the memory. The memo recording unit 32 has memo recordings (memo recording areas) 32a to 32c.
[0135]
Thereby, the processors 22a to 22e provided in each communication device, the ROM 23 and the RAM 24 (see FIG. 4) cooperate to function as the call processing units 2a to 2e. Then, the resource status monitoring units (not shown) of these call processing units 2a to 2e monitor the resource status obtained in the shared memory 29. This makes it possible to check, for example, the operation state of software having a call processing function for each of the call processing units 2a to 2e.
[0136]
Further, when each of the call processing units 2a to 2e recognizes an abnormality in the resource state, the call processing unit 2a restores the shared memory 29 using the state monitoring table, and includes a management unit and a recovery unit (not shown). It is configured.
Here, the management unit divides the resources of the shared memory 29 into pages of a predetermined size, manages whether or not each page has been updated, retains the information in a management data file, and updates the data of the updated page. In the update data file.
[0137]
Further, the recovery unit recovers data held in the shared memory 29 based on information of the management data file and the update data file at the time of failure recovery. Then, for example, the call processing unit 2a and the shared memory 29 shown in FIG. 17 cooperate to perform the functions of the management unit and the recovery unit.
With this management unit, it is possible to ascertain the status of a device (device-device or device-terminal) corresponding to the fault occurrence portion of the status monitoring table for the resource (i.e., local fault occurrence status).
[0138]
The state monitoring table 30 is connected to each of the recovery unit 33, the initialization unit 34, and the selection unit 35, and is referred to by the recovery unit 33, the initialization unit 34, and the selection unit 35. I have.
Here, the recovery unit 33 recovers a local failure portion of the state monitoring table 30 based on the communication state held in the memo recording unit 32. The initialization unit 34 initializes a local failure occurrence part among the plurality of communication states 1 to N held by the state monitoring table 30. Further, the selection unit 35 implements one of the recovery unit 33 and the initialization unit 34.
[0139]
Therefore, the state monitoring table 30 does not hold all the information in the management data file in the recovery process after the occurrence of the failure, but initializes the table contents and invalidates the information. It is possible to perform two local recovery processes including a recovery process method in which only information is recorded and the state is reset. As a result, the status of a device such as the BTS 83p (see FIG. 17) or the resource of the MS 10a is monitored or managed using the status monitoring table 30. More specifically, in the mobile communication system 100a, when the partner terminal 11a is talking with the MS 10a, the call between the partner terminal 11a and the MS 10a is transmitted through the public network 85a, the core network 101, the exchange 102, the RNCs 84a to 84b, and They are connected via BTSs 83a to 83c. In a state where the call is connected, each network or various communication devices mutually manage and monitor the ID and the like of the communication partner device.
[0140]
This also eliminates the need for the operator of the monitoring system 100a or the fault occurrence unit of the management system 100a to constantly check the queue in the shared memory 29 of the RNC 84p. Therefore, it takes time and man-hours to elucidate, and the occurrence of a failure can be found relatively early.
With such a configuration, when a failure occurs, the communication state is invalidated by recovering the state monitoring table 30 (first recovery) based on the memo record and initializing the recovered state monitoring table 30. Recovery (second recovery) is selectively performed.
[0141]
As a result, it is not necessary to manage the entire mobile communication system 100a, and a failure that has occurred between local communication devices can be recovered by referring to the state monitoring table 30.
As described above, when various communication devices are managed using the state monitoring table 30, even if a failure occurs at the time of accessing the table by the processors 22a to 22e, the table is not destroyed. In addition, both abnormal operation of the program and occurrence of a processing error of the system due to the occurrence of the fault are avoided.
[0142]
The state monitoring system according to the second embodiment can be applied to a system using a multiprocessor system, electronic equipment, and the like, and can be applied to various units such as a terminal, a call, a base station, and a computer.
An example in which this monitoring system is applied to a mobile communication system 100 (see FIG. 2) will be described. The monitoring system can be used not only for mobile communication but also for other uses.
[0143]
(C) Other
The present invention is not limited to the above-described embodiment and its modifications, and can be implemented with various modifications without departing from the spirit of the present invention.
(D) Additional notes
(Supplementary Note 1) In a multiprocessor system having a plurality of processors for processing a communication call and a shared memory accessible by the plurality of processors,
The shared memory is
A queue management information table holding queue management information about a resource management queue holding call information related to the communication call;
An update history recording area for recording an update history of the queue management information of the queue management information table.
[0144]
(Supplementary note 2) In a multiprocessor system having a plurality of processors for processing a communication call and a shared memory accessible by the plurality of processors,
The shared memory is
A queue management information table holding queue management information including a plurality of information elements for specifying the size of the resource management queue holding the call information related to the communication call acquired by each processor and the position of the queue;
An update history recording area for recording an update history composed of a plurality of update information elements to be written in the queue management information of the queue management information table.
[0145]
(Supplementary note 3) In a multiprocessor system having a plurality of processors for processing a communication call and a shared memory accessible by the plurality of processors,
The shared memory is
A queue management information table holding queue management information including at least a queue management number and a head queue address for a resource management queue holding call information related to the communication call acquired by each processor;
A multi-processor system provided with an update history recording area for recording an update history including a plurality of update information elements to be written to the queue management number held in the queue management information table and the head queue address. .
[0146]
(Supplementary Note 4) When a failure occurs when any of the plurality of processors accesses the queue management information, the queue management information stored in the queue management information table and the update history record are stored. A determination unit that determines whether recovery processing is necessary based on the update history information recorded in the area,
Any one of supplementary notes 1 to 3, further comprising a recovery section for recovering the queue management information based on the update history information when the determination section determines that recovery is necessary. 2. The multiprocessor system according to 1.
[0147]
(Supplementary Note 5) A plurality of processors that process communication calls, a shared memory that can be accessed by the plurality of processors, and a queue management information table that holds queue management information about a resource management queue that holds call information about the communication calls. In a multiprocessor system having
The shared memory is
An update history recording area for recording an update history of the queue management information of the queue management information table.
[0148]
(Supplementary note 6) In a multiprocessor system having a plurality of processors for processing a communication call and a shared memory accessible by the plurality of processors,
Each processor
A queue management information table holding queue management information about a resource management queue holding call information related to the communication call;
A multiprocessor system characterized by sharing an update history recording area for recording an update history of queue management information in the queue management information table.
[0149]
(Supplementary Note 7) In a multiprocessor system having a plurality of processors for processing information data and a shared memory accessible by the plurality of processors,
The shared memory is
A queue management information table that holds queue management information about a resource management queue that holds various information related to the information data,
An update history recording area for recording an update history of the queue management information of the queue management information table.
[0150]
(Supplementary Note 8) A plurality of processors for processing the communication call;
A memory having a queue management information table holding queue management information on a resource management queue accessed by the plurality of processors and holding call information on the communication call;
Each processor
A multiprocessor system, characterized by sharing an update history recording area provided in the memory for recording an update history of queue management information in the queue management information table.
[0151]
(Supplementary Note 9) In a multiprocessor system having a plurality of processors that process call information related to communication and a shared memory that can be accessed by the plurality of processors and hold the call information,
The shared memory is
A queue management information table holding queue management information about a resource management queue holding call information related to the communication call;
An update history recording area for recording an update history of the queue management information of the queue management information table.
[0152]
(Supplementary Note 10) A plurality of processors for processing the communication call;
A shared memory accessible by the plurality of processors;
The shared memory is
A queue management information table holding queue management information about a resource management queue holding call information on the communication call;
An update history recording area for recording an update history of the queue management information of the queue management information table.
[0153]
(Supplementary Note 11) In a monitoring system that monitors a communication state between a communication device and an opposite communication device that communicates with the communication device,
A state monitoring table for holding the communication state;
A communication state recording area for recording the communication state held in the state monitoring table;
A recovery unit that recovers a local failure occurrence portion of the status monitoring table based on the communication status held in the communication status recording area;
A monitoring system, comprising: a selection unit that performs at least one of an initialization unit that initializes a local failure occurrence part in a communication state of the state monitoring table.
[0154]
(Supplementary Note 12) A plurality of processors for processing the communication call;
A distributed control unit that distributes a communication call from an external device according to a load of the plurality of processors;
A shared memory accessible by the plurality of processors;
The shared memory is
A queue management information table holding queue management information about a resource management queue holding call information related to the communication call;
An update history recording area for recording an update history of the queue management information in the queue management information table.
[0155]
(Supplementary Note 13) A queue retrieval method in a multiprocessor system having a plurality of processors for processing a communication call and a shared memory accessible by the plurality of processors,
A reference step of referring to a queue management information table holding queue management information including a plurality of information elements for specifying the size of the resource management queue and the position of the resource management queue acquired by any one of the plurality of processors; ,
A recording step in which each processor records an update history including a plurality of update information elements to be written in the queue management information of the queue management information table in an update history recording area;
A writing step of writing the update history into the queue management information table by each processor; a queue retrieval method in a multiprocessor system.
[0156]
(Supplementary Note 14) In the referring step, each processor refers to at least one of a queue management number and a head queue address as the plurality of information elements,
In the recording step, each processor records, as the update information element, at least one of a first update value of the queue management number and a second update value of the head queue address,
14. The queue retrieval method according to claim 13, wherein in the writing step, each processor writes at least one of the first update value and the second update value in the queue management information table.
[0157]
(Supplementary Note 15) A table recovery processing method in a multiprocessor system having a plurality of processors for processing a communication call and a shared memory accessible by the plurality of processors,
The queue management information of the queue management information table holding queue management information including a plurality of information elements for specifying the size of the resource management queue acquired by any one of the plurality of processors and the position of the queue. A determination step of determining whether recovery processing is necessary based on the update history information in an update history recording area that records an update history composed of a plurality of update information elements to be written in the queue management information of the queue management information table;
A recovery step of recovering the queue management information based on the update history information when it is determined that the recovery is necessary in the determination step; a table recovery processing method in a multiprocessor system.
[0158]
(Supplementary Note 16) In the determining step, each processor may determine whether the first queue management number or the first head queue address as a plurality of information elements of the queue management information and the second queue management as the plurality of update information elements. The necessity is determined based on the number or the match / mismatch with the second head queue address;
The recovery step is characterized in that each of the processors rewrites the second queue management number or the second head queue address to the first queue management number or the first head queue address based on the necessity. The table recovery processing method according to supplementary note 15.
[0159]
【The invention's effect】
As described in detail above, the multiprocessor system (claims 1 and 2), the monitoring system (claim 3), the queue retrieval method in the multiprocessor system (claim 4), and the table recovery process in the multiprocessor system according to the present invention. According to the method (claim 5), the following effects or advantages can be obtained.
[0160]
(1) According to the multiprocessor system of the present invention, in a multiprocessor system having a plurality of processors for processing a communication call and a shared memory accessible by the plurality of processors, the shared memory holds call information on the communication call. A queue management information table that holds queue management information about resource management queues to be managed and an update history recording area that records an update history of queue management information in the queue management information table are provided, so that processing is simplified and high speed is achieved. , And a highly reliable multiprocessor system is realized.
[0161]
(2) According to the multiprocessor system of the present invention, the shared memory is composed of a plurality of information elements for specifying the size and the position of the resource management queue for holding the call information related to the communication call obtained by each processor. Since a queue management information table that holds queue management information and an update history recording area that records an update history composed of a plurality of update information elements to be written in the queue management information of the queue management information table are provided, the queue management information table is When the legitimacy is not established due to a failure, it can be recovered automatically, quickly and completely (claim 1).
[0162]
(3) According to the multiprocessor system of the present invention, the queue management information including at least the queue management number and the head queue address of the resource management queue that holds the call information related to the communication call acquired by each processor. And a queue management information table for holding the queue management information table and an update history recording area for recording an update history composed of a plurality of update information elements to be written to the queue management number and the first queue address held in the queue management information table. Regarding the processing performance at the time of memo recording and recovery, the queues enqueued in the queue management information table are searched from both the first queue and the last queue, and the search time is longer depending on the number of queues and the search time is variable. For the method, the number of queue management in the queue management information table in the memo record, or Since the head queue address information is retained, there is no need to search for whether the connection to the queue is normal or abnormal, and the memo recording can use a minimum area to determine the presence or absence of recovery processing, making effective use of shared memory. Can be achieved.
[0163]
(4) When a failure occurs during access to the queue management information by each processor, based on the queue management information held in the queue management information table and the update history information recorded in the update history recording area, It may be configured to include a determination unit that determines whether recovery processing is necessary, and a recovery unit that recovers queue management information based on update history information when the determination unit determines that recovery is necessary. In the case where the number of received signals is increased and the search time is required to the maximum with respect to the system reliability, a double failure or an exclusive acquisition failure by another processor is induced. Can be performed in a single time and can be performed at high speed, so that induction of a double failure is avoided (claim 2).
[0164]
(5) According to the multiprocessor system of the present invention, in a multiprocessor system having a plurality of processors, a shared memory, and a queue management information table, the shared memory is provided with the update history recording area. Avoid errors.
(6) According to the multiprocessor system of the present invention, in a multiprocessor system having a plurality of processors, a shared memory, and a queue management information table, each processor shares the queue management information table and the update history recording area. In addition, the processing performance due to a call loss at the time of recovery can be eliminated, and another processor can access when a failure occurs while accessing the shared memory by the processor.
[0165]
(7) According to the multiprocessor system of the present invention, in a multiprocessor system including a plurality of processors that process information data and a shared memory that can be accessed by the plurality of processors, the shared memory holds various information related to the information data. A queue management information table for holding queue management information about a resource management queue to be executed and an update history recording area for recording an update history of the queue management information in the queue management information table are provided. However, simplification of processing, reliability using a large amount of program capacity, and flexibility due to allocation of a relatively large working memory capacity can be obtained.
[0166]
(8) According to the multiprocessor system of the present invention, a plurality of processors and a memory having a queue management information table are provided, and each processor shares an update history recording area. Recovery processing can be performed only by writing information to the memo record.
(9) According to the multiprocessor system of the present invention, the multiprocessor system has a plurality of processors for processing call information related to communication and a shared memory which can be accessed by the plurality of processors and hold the call information, and the shared memory is a queue management information table. And the update history recording area, the recovery process can be simplified.
[0167]
(10) According to the multiprocessor system of the present invention, a plurality of processors and a shared memory are provided, and the shared memory is provided with the queue management information table and the update history recording area. It becomes possible at high speed.
(11) According to the multiprocessor system of the present invention, in a monitoring system that monitors a communication state between a communication device and an opposing communication device that communicates with the communication device, a status monitoring table that holds the communication status, A communication state recording area for recording the communication state held in the table, a recovery unit for recovering a local failure portion of the state monitoring table based on the communication state held in the communication state recording area, and a state monitor Since there is provided an initializing unit for initializing a local failure occurrence portion of the communication state of the table and a selecting unit for implementing at least one of the initializing unit, a local recovery processing method becomes possible and the amount of memory is reduced. There is a fail-safe action that reduces the speed, increases the speed, and does not cause the system to go down.
[0168]
(12) According to the load distribution device of the present invention, a plurality of processors that process communication calls, a distribution control unit that distributes communication calls from external devices according to the loads of the plurality of processors, The shared memory has a queue management information table that holds queue management information about a resource management queue that holds call information related to a communication call, and an update history of the queue management information in the queue management information table. Since the update history recording area is provided, the induction of a double failure is avoided.
[0169]
(13) According to the queue retrieval method in the multiprocessor system of the present invention, a queue including a plurality of information elements for specifying the size and the position of the resource management queue acquired by any of the plurality of processors. A reference step of referring to a queue management information table holding management information; and a recording step of recording, in an update history recording area, an update history including a plurality of update information elements written in the queue management information of the queue management information table by each processor. And each processor has a writing step of writing the update history to the queue management information table, so that the recovery process can be performed only by recording a memo in the queue management information table. Recovery is possible, simplifying the system (Claim 4).
[0170]
(14) In the referring step, each processor refers to at least one of a queue management number and a head queue address as a plurality of information elements, and in the recording step, each processor includes a queue management information as an update information element. Recording at least one of the first update value of the first number and the second update value of the first queue address, and the writing step includes the steps of: causing each processor to queue at least one of the first update value and the second update value; The information may be written in the management information table. In this way, the burden of maintenance and the like is reduced, and it is possible to flexibly respond to changes in system specifications.
[0171]
(15) According to the table recovery processing method in the multiprocessor system of the present invention, the table recovery processing method includes a plurality of information elements for specifying the size and the position of the resource management queue acquired by any of the plurality of processors. On the basis of the queue management information of the queue management information table holding the queue management information and the update history information of the update history recording area for recording the update history composed of a plurality of update information elements written in the queue management information of the queue management information table Since it has a judgment step of judging the necessity of the recovery process and a recovery step of recovering the queue management information based on the update history information when the judgment step judges that the recovery is necessary, the local recovery Processing method becomes possible, reducing memory amount, high speed, and system down There is a fail-safe action is not caused.
[0172]
(16) In the determining step, each processor determines whether the first queue management number or the first head queue address as a plurality of information elements of the queue management information and the second queue management number or the second head queue as a plurality of update information elements. The necessity is determined based on the match / mismatch with the queue address. In the recovery step, each processor determines the second queue management number or the second head queue address based on the necessity or not by the first queue management number or the first queue management number. Since the first queue address is rewritten, there is no need to manage the entire system, and a failure recovery process between local communication devices of the mobile communication system can be performed.
[0173]
Also, when the status monitoring is managed using a table, or when a failure occurs at the time of accessing the table, the table is not destroyed, and a system processing error is avoided.
[Brief description of the drawings]
FIG. 1 is a configuration diagram of a mobile communication system according to a first embodiment of the present invention.
FIG. 2 is a functional block diagram of the mobile communication system according to the first embodiment of the present invention.
FIG. 3 is a schematic block diagram of a base station control device according to the first embodiment of the present invention.
FIG. 4 is a diagram showing a main part of the load distribution device according to the first embodiment of the present invention.
FIG. 5 is a diagram schematically showing an area of a shared memory according to the first embodiment of the present invention.
FIG. 6 is a schematic view of the load distribution device according to the first embodiment of the present invention as viewed from above.
FIG. 7 is a diagram for explaining a recovery process according to the first embodiment of the present invention.
FIG. 8 is a diagram for explaining a head queue retrieval process according to the first embodiment of the present invention.
FIG. 9 is a flowchart for explaining a head queue removal process according to the first embodiment of the present invention.
FIG. 10 is a flowchart illustrating a recovery process for removing a head queue according to the first embodiment of the present invention.
FIG. 11 is a diagram for explaining a method of taking out an intermediate queue according to the first embodiment of the present invention.
FIG. 12 is a flowchart illustrating a process of taking out a halfway queue according to the first embodiment of the present invention.
FIG. 13 is a flowchart illustrating a recovery process for removing a halfway queue according to the first embodiment of the present invention.
FIG. 14 is a diagram for explaining an enqueue method according to the first embodiment of the present invention.
FIG. 15 is a flowchart illustrating an enqueue process according to the first embodiment of the present invention.
FIG. 16 is a flowchart illustrating an enqueue recovery process according to the first embodiment of the present invention.
FIG. 17 is a configuration diagram of a mobile communication system according to a second embodiment of the present invention.
FIG. 18 is a diagram schematically illustrating an area of a shared memory according to a second embodiment of the present invention.
FIG. 19 is a block diagram of a load distribution type multiprocessor system.
FIG. 20 is a diagram showing a sequence for explaining resource acquisition from a shared memory.
21A is a diagram illustrating an example of shared memory allocation before a failure occurs, and FIG. 21B is a diagram illustrating an example of shared memory allocation after a failure occurs.
FIG. 22 is a diagram illustrating an example of a recovery processing operation of the shared memory.
FIGS. 23A to 23F are diagrams for explaining dequeue or enqueue; FIG.
[Explanation of symbols]
1 load balancer
2a to 2e Call processing unit
3,29 shared memory
4 Hard disk drive
5 Communication device between processors
5e bus
5f debugger
6a-6c, 6q resource (queue)
6d table area
8,32 memo recording area (update history recording area)
9.1a Distributed control unit
10a, 10p mobile phone
10b video playback device
10c videophone
11a, 11b Communication partner terminal (partner terminal)
22a to 22e processors
23 ROM
24 RAM
25a judgment unit
25b Recovery unit
32a-32c memo record
33 Recovery unit
34 Initialization unit
35 Selector
82 Backboard
83a-83c, 83p BTS (wireless base station)
84a-84b, 84p RNC (Radio Network Controller)
85a line interface
85b ATM switch
85c output interface
85d terminal
86a Voice receiver
86g voice transmitter
86b Baseband processing unit
86c modem
86d transmission / reception amplifier
86e key information storage
86f, 86p mobile unit controller
87a Highway in control unit
87b, 87p base station controller
88 Clock section
88a power supply
89a DHT
89b M-MUX
100,100a Mobile communication system

Claims (5)

In a multiprocessor system having a plurality of processors for processing communication calls and a shared memory accessible by the plurality of processors,
The shared memory is
A queue management information table holding queue management information including a plurality of information elements for specifying the size of the resource management queue holding the call information related to the communication call acquired by each processor and the position of the queue;
An update history recording area for recording an update history composed of a plurality of update information elements to be written in the queue management information of the queue management information table. When any one of the plurality of processors fails when accessing the queue management information, the queue management information stored in the queue management information table and the queue management information stored in the update history recording area are recorded. A determination unit that determines whether recovery processing is necessary based on the updated history information,
2. The multiprocessor system according to claim 1, further comprising a recovery unit that recovers the queue management information based on the update history information when the determination unit determines that recovery is necessary. In a monitoring system that monitors a communication state between a communication device and an opposite communication device that communicates with the communication device,
A state monitoring table holding the communication state including the resource state of the shared memory accessible by a plurality of processors,
A communication state recording area for recording the communication state held in the state monitoring table;
A recovery unit for recovering a local failure portion of the status monitoring table based on the communication status held in the communication status recording area; and a local failure occurrence portion of the communication status of the status monitoring table. A selection to perform at least one of initialization and initialization of the monitoring system. A queue retrieval method in a multiprocessor system including a plurality of processors for processing a communication call and a shared memory accessible by the plurality of processors,
A reference step in which one of the plurality of processors references a queue management information table holding queue management information including a plurality of information elements for specifying the size of the acquired resource management queue and the position of the queue; When,
A recording step in which any of the processors records an update history including a plurality of update information elements to be written in the queue management information of the queue management information table in an update history recording area;
A writing step of writing the update history to the queue management information table by any one of the processors. A table recovery processing method in a multiprocessor system having a plurality of processors for processing a communication call and a shared memory accessible by the plurality of processors,
The queue management information of the queue management information table holding queue management information including a plurality of information elements for specifying the size of the resource management queue acquired by any one of the plurality of processors and the position of the queue. A determination step of determining whether recovery processing is necessary based on the update history information in an update history recording area that records an update history composed of a plurality of update information elements to be written in the queue management information of the queue management information table;
A recovery step of recovering the queue management information based on the update history information when it is determined that the recovery is necessary in the determination step; a table recovery processing method in a multiprocessor system.

Images