JPH0677997A - Load distributing method for packet exchange - Google Patents

Abstract

(57) [Abstract] [Objective] The present invention relates to a load distribution method for a multiprocessor packet switch, and aims to distribute the load at the time of call setup / release processing among a plurality of processors. [Constitution] Congestion state management means 11 that stores one or a plurality of lines 3 and stores the congestion state of each processor in each of a plurality of processors 1 _-1 to 1 _-n connected via a bus 2. , When there is a call from the accommodated line, the congestion status of each processor is confirmed by the congestion status management means. If the own processor is not in the congestion status, the own processor is not. If the own processor is in the congestion status, it is not in the congestion status. Call setting processing processor selecting means 12 for selecting a processor and requesting execution of call setting / cancelling processing, and call setting when call setting / cancelling processing is requested by the processor selecting means for executing call setting processing or another processor A call setup / cancellation process executing means 13 for executing the / cancellation process is provided.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a load distribution method for a multiprocessor type packet switch. 2. Description of the Related Art In recent years, a multiprocessor type information processing system or a switching system in which a plurality of processors share the processing in order to improve the processing capacity has been widely used. In the example of the multiprocessor type packet switch, a plurality of processors accommodates one or a plurality of lines, respectively, and a calling line and a called line are configured to perform packet communication through the processors which accommodate the respective lines. There are many things.

Packet communication processing in a packet switch is a call setting and call release processing (hereinafter called call processing) that performs routing processing, service function setting processing (usually called facility processing), billing processing, etc. at the start and end of communication. The setting process and the call release process are collectively referred to as a call setup / cancellation process) and the flow control process for controlling the transfer of data packets. It usually has more throughput.

In order to set up / release a call, a lot of necessary data (hereinafter referred to as reference data) such as subscriber (terminal) data and many processing steps are required. Therefore, the packet switch specifies the call setup / release processing. Many processors are configured to perform call setup, but there is also a configuration in which a plurality of processors accommodating lines perform call setup / cancellation processing for calls from lines accommodated in their own processors. .

On the other hand, a user of a packet switch often uses a packet communication network constituted by the packet switch to perform packet communication between a large number of terminals and a small number of host machines. In many cases, the host machine performs control communication with a large number of terminals when the host machine is activated or deactivated at the beginning or end of work. In such a time zone, there occurs a phenomenon in which call setup / cancellation processing extremely increases as compared with communication processing for data packet transfer.

In the packet switch configured so that the call setup / cancellation process is performed by a specific processor, it is possible to specify a specific processor although the load of a plurality of processors accommodating the line is small during the above-mentioned time zone. Often only the processor is overloaded.

On the other hand, in a packet switch having a configuration in which a plurality of processors accommodating lines perform call setup / cancellation processing for calls originating in the accommodating line of the own processor, the load of the processors is dispersed to some extent. However, in the time zone as described above, only the processor accommodating the host machine is often overloaded due to the call setup / release processing, and the load distribution is not uniform.

From the above, in the multiprocessor packet switch, there is a need for a load distribution method in which the load involved in call setup / cancellation processing is not concentrated on a particular processor.

[0008]

2. Description of the Related Art FIGS. 4 and 5 show packet switching in the prior art.
The structural example of the machine is shown. FIG. 4 shows a management processor (hereinafter
(Hereinafter referred to as MPR) 41 concentrates on call setup / release processing.
However, in this configuration, 1 or
Are multiple processors 31 that accommodate multiple lines_-1~ 31_-nIs
Mainly performs flow control processing. The illustration shows multiple processors.
As a processor 31_-1And 31_-2Only two of them are shown, both
Processor 31 _-1, 31_-23 lines of 10 lines each_-Ten
~ 3_-19And line 3_-20~ 3_-29Is housed
Is shown.

A terminal is connected to each line.
In the terminal 4 _{-1 0} to 4 _-19 to line 3 _-10 -3 _-19 is, a state in which the terminal 4 _-20 -4 _-29 to line 3 _-20 -3 _-29 are connected are shown ( (Only lines and terminals are shown).
These terminals _4-10 to _4-29 are _DTs in data communication.
The terminal device includes a terminal device called E, a line terminating device called DCE, and the like. The terminal device includes a pure terminal, a personal computer, a workstation, or a host computer.

Hereinafter, the processor 31 _-1 (hereinafter, CPR # 1
And referred) to stowed terminal 4 _-10 (hereinafter, terminal T referred to as ₁₀₎ from the processor 31 ₂ (hereinafter, CPR # 2 and referred) to accommodated the terminal 4 _-20 (hereinafter referred to as the terminal T ₂₀ ) Is called and packet communication is performed between both terminals, the processing of the packet switch will be described as an example.

A call request packet (hereinafter referred to as a CR packet) for the terminal T ₁₀ to perform packet communication with the terminal T _20.
Is transmitted, the processing unit 30 of the CPR # 1 receives the CR packet via the input / output interface unit 37, and transfers the CR packet from the bus interface unit 36 to the MPR 41 via the bus 2. The CR packet includes the subscriber number of the calling terminal, but the call setup / cancellation processing execution unit 43 of the MPR 41 receives the CR packet, accesses the reference data 42, and the called terminal T ₂₀ performs CPR. The content of the CR packet is notified to CPR # 2 by identifying that it is accommodated in # 2. At that time, the calling processor makes CPR # 1.
Include information that identifies

[0012] CPR # 2 receives the above notification, the incoming call packet (hereinafter, referred to as CN packet) to the called terminal T ₂₀ a letter. When the called terminal T ₂₀ returns an incoming call acceptance packet (hereinafter referred to as a CA packet) informing that the incoming call is accepted, CPR # 2 transfers this to the MPR 41, and the call setup / cancellation processing execution unit of the MPR 41. 43 CP this
Transfer to R # 1, but at this time, notify that the processor accommodating the called terminal T ₂₀ is CPR # 2.

[0013] CPR # 1 Upon receiving a notification of incoming call reception to the terminal T _10, connection completion packet indicating that the connection to the partner terminal is completed (hereinafter, referred to as CC packet)
To send. This completes the call setup process, but the MPR
The call setup / cancellation process execution unit 43 of 41 stores the information about the communication between the terminals T ₁₀ and T ₂₀ in a call control memory (not shown) until the call cancellation process is completed, and manages the call. .

When the calling terminal T ₁₀ receives the CC packet, transmission / reception of the data packet is started between the terminals T ₁₀ and T ₂₀ , but the data packet is transferred by the CPR.
This is performed by the flow control process of the processing unit 30 of # 1 and CPR # 2.

When the packet communication ends and, for example, the terminal T ₁₀ transmits a restoration request packet (hereinafter referred to as a CQ packet) indicating the end of the call, the processing unit 30 of CPR # 1 receives this and transfers it to the MPR 41. To do. MPR41 call setup /
Upon receiving this, the cancellation processing execution unit 43 notifies the CPR # 2 that there is a restoration request, and the processing unit of the CPR # 2
30 sends a disconnection instruction packet (hereinafter referred to as a CI packet) to the terminal T ₂₀ .

Upon confirming that the call is to be disconnected, the terminal T ₂₀ sends a disconnection confirmation packet (hereinafter referred to as a CF packet) to CPR # 2, and CPR # 2 transmits this CF packet to MPR.
Transfer to PR41.

When the MPR 41 receives the CF packet, C
The amount of packets transmitted from PR # 1 to CPR # 2 (or the amount of packets received from terminal T ₁₀ ) and the amount of packets transmitted from CPR # 2 to terminal T ₂₀ are received from CPR # 1 and CPR # 2. Then, the call cancellation process is started, for example, the billing process is performed after matching.

When the billing process is completed, the MPR 41 displays C
The contents of the CF packet are sent to PR # 1 to notify the disconnection of the call, the call control memory is released, and the call management is terminated. When CPR # 1 receives the notification, the restoration confirmation packet (hereinafter, referred to as CF packet) to verify that the recovery to the terminal T ₁₀ is finished sending, the packet communication between the terminal T ₁₀ and the terminal T ₂₀ ends To do. In the above,
The call cancellation process is from the reception of the CQ packet from the calling terminal to the transmission of the CF packet to the terminal.

As is apparent from the above description, call setup /
Most of the release processing is the call setup / release processing execution unit of MPR41
The CP at the time of call setup and call release performed by 43
The processing of R # 1 and CPR # 2 is performed by the terminal T ₁₀ or the terminal T _20.
The main processing is the relay processing of call setup / cancellation related packets between the MPR41 and the MPR41. The main processing of CPR # 1 and CPR # 2 is the flow control processing, but the processing amount of the call setup / cancellation process and the processing amount of the flow control in packet communication often have a ratio of about 10: 1, for example. When calls are concentrated at the start of work, only the MPR 41 is often in a congestion state.

FIG. 5 shows call setup / cancellation processing performed by a plurality of processes.
It shows the configuration shared by the service provider. In this configuration each
Processing of multiple processors accommodating one or more lines
Part is equipped with a call setup / cancellation process execution part and accommodates the calling terminal
The processor that executes the call setup / release processing. MPR
(Management processor) 61 does not perform call setup / release processing
However, the common data required for call setup / cancellation is referred to as reference data 62.
To prepare. In addition, in FIG.
Each line 3_-Ten~ 3_-19And line 3_-20~ 3 _-29
Two processors 51 containing_-1And 51_-2Only illustrated
There is.

Hereinafter, the processor 51 _-1 (hereinafter, CPR # 1
An example in which a call is made to a terminal T ₂₀ accommodated in a processor 51 _-2 (hereinafter referred to as CPR # 2) from a terminal T ₁₀ accommodated in 5 will be described with reference to FIG. 5, but the portions overlapping with FIG. 4 will be briefly described.

When the terminal T ₁₀ sends a CR packet for packet communication with the terminal T ₂₀ and the processing unit 50 of CPR # 1 receives the CR packet, the call setup / call setup in the processing unit 50 is performed.
The cancellation processing execution unit 53 starts the call setting processing. The call setup / cancellation processing execution unit 53 first accesses the reference data 63 of the MPR 61 via the bus 2, and the call receiving terminal T ₂₀ is accommodated in the C.
Confirm PR etc. As a result, the terminal T ₂₀ becomes CPR #
When it is confirmed that the data is accommodated in the No. 2, the contents of the CR packet are notified to the CPR # 2 via the bus 2.

[0023] CPR # 2 receives the notice, sends the CN packet to the terminal T _20, the terminal T ₂₀ is the return of the CA packet, and transfers its contents to the CPR # 1.
When the CPR # 1 receives the contents of the CA packet, the terminal T ₁₀
A CC packet is sent to the terminal to terminate the call setup process. At this time, information about communication between the terminals T ₁₀ and T ₂₀ is stored in a call control memory (not shown) in the call setup / cancellation process execution unit 53. Is stored and the call is managed until the call release processing is completed.

When the calling terminal T ₁₀ receives the CC packet, transmission / reception of the data packet is started between the terminals T ₁₀ and T ₂₀ , but the data packet is transferred by the CPR.
This is performed by the flow control processing of the processing unit 50 of # 1 and CPR # 2.

When the packet communication is completed and the terminal T ₁₀ transmits the CQ packet, the call setting / processing of the processing unit 50 of the CPR # 1 is performed.
The cancellation processing execution unit 53 sends the contents to CPR # 2, and the CP
The call setting / cancelling process executing unit 53 of the processing unit 50 of R # 2 is the terminal T
Send CI packet to ₂₀ . When the terminal T ₂₀ confirms the disconnection of the call and sends the CF packet to the CPR # 2, the CPR # 2 transfers the content of the CF packet to the CPR # 1.

When the CPR # 1 receives the contents of the CF packet, the CPR # 2 receives from the CPR # 2 information regarding the packet amount transmitted to the terminal T ₂₀ , and the CPR # 1 transmits to the CPR # 2. Packet volume (or terminal T
_The amount of packets received from ₁₀ ) is checked and billing is performed. When the charging process is completed, CPR # 1 sends CF to terminal T ₁₀ .
Send a packet and release the call control memory,
The call release processing ends.

As described above, in the configuration shown in FIG. 5, the call setup / cancellation processing is performed by the call setup / release of the CPR # 1 accommodating the calling terminal T _10.
Since the call release / execution process is performed by the release processing execution unit 53, the call setup / release processing is rarely concentrated on a specific processor. For example, when the terminal T ₁₀ is a host computer and all terminals are controlled at the start of work. When communication for communication is performed, a phenomenon occurs in which only CPR # 1 is in a congestion state.

[0028]

Since the processing amount of the call setup / cancellation process in packet communication is extremely large compared to the processing amount of the flow control, the call setup / cancellation process is configured to be performed by a specific processor. In addition, the conventional multiprocessor packet switch has a drawback that only a specific processor that performs a call setup / cancellation process is often in a congestion state. Further, in a conventional multiprocessor packet switch configured so that a plurality of processors accommodating lines share the call setup / cancellation processing, only the processor accommodating the host computer is congested at the start of work. Is likely to occur. That is,
The conventional multiprocessor packet switch has a problem that the processing capacity of the switch is restricted by the occurrence of such congestion.

An object of the present invention is to distribute the load at the time of call setup / release processing among a plurality of processors.

[0030]

FIG. 1 illustrates the principle of the present invention.
It is a figure. Figure 1 shows a multiprocessor packet switch
Shows the configuration of_-1Through 1 _-nAre 1 or
A plurality of lines for accommodating a plurality of lines 3 and processing packet communication
Processors 2 are the plurality of processors 1_-1~ 1_-nContact
It is a bus to be continued.

Reference numerals 11 to 13 denote the plurality of processors 1 _-1 to 1
_-means provided in _n , 11 monitors the congestion state of its own processor, stores it as congestion state information and sends the congestion state information to other processors, and the congestion state information sent from other processors Congestion state management means for receiving and storing.

Reference numeral 12 confirms the congestion state of each processor by the congestion state management means 11 when a call is made from the line 3 to be accommodated, and when the own processor is not in the congestion state, call setting / cancellation processing of the own processor. A call setting processor that requests the executing means 13 to execute call setting / cancellation processing, and when the processor itself is in a congested state, selects one of the processors not in the congested state and requests the execution of the call setting / cancellation processing. It is a means of selection.

Reference numeral 13 designates a call setup / cancellation process when the call setup / execution processor selecting means 12 requests the execution of the call setup / cancellation process, or when another processor requests the execution of the call setup / cancellation process. Is a call setup / cancellation process executing means for executing.

[0034]

In FIG. 1, the congestion state management means 11 of each of the processors 1 _-1 to 1 _-n monitors the congestion state of its own processor, stores it as congestion state information, and stores this congestion state information in other processors. Sending out. When the congestion state information of the processor is sent from another processor, the congestion state information is received and stored together with the congestion state information of the own processor.

The processor, for example when there is a call from the line 3 to the processor 1 _-1 is accommodated, the call setting processing processor selecting means 12 to confirm the congestion status of each processor by congestion state management unit 11, the own processor When it is not in the congestion state, it requests the call setup / release processing executing means 13 of its own processor to execute the call setup / release processing. When the call setting / releasing process executing means 13 is requested by the call setting process executing processor selecting means 12 to execute the call setting / releasing process, it executes the call setting / releasing process.

When the call setting processing processor selecting means 12 confirms the congestion state of each processor, and if its own processor is in the congestion state, the call setting processing processor selecting means 12 selects one of the processors not in the congestion state, The selected processor is requested to execute the call setup / cancellation processing via the bus 2. In the requested processor, the call setup / cancellation processing execution means 13 executes the call setup / cancellation processing of the call.

As described above, in the configuration of FIG. 1, even when calls are concentrated from the line 3 accommodated in a specific processor, a plurality of processors share the call setup / cancellation processing. Therefore, only the specific processor is not congested, and the processing capability of the packet switch is not restricted by the congestion of the specific processor.

[0038]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIGS. 2 and 3 are block diagrams of an embodiment of the present invention. Throughout the drawings, the same reference numerals indicate the same object, and 1a _-1 ,
1a _-2 , 1b _-1 , 1b _-2 are processors, 2 are buses, 3 _{-10 to} 3
_-29 is processor 1a _-1 , 1a _-2 or processor 1b _-1 , 1b
_Lines accommodated in _-2 , 4 _-10 to 4 _-19, 4 _-20 to 4 _-29
Each line _{3-1 0-3 -19,} a terminal connected to the 3 _-20 -3 _-29.

The processors 10a, 10b and 11 to 17 are processors 1a _-1 , 1a.
_-2 and each part provided in the processor 1b _-1 , 1b _-2 , 10
a is a processing unit of processors 1a _-1 and 1a _-2 , and 10b is a processor
1b _-1 , 1b _-2 processing unit, 11 is a congestion state management unit, 14 and 15 are congestion state storage table and congestion state information transmission and reception unit constituting the congestion state management unit 11, 12 is a call setting processor selection unit, Reference numeral 13 is a call setup / cancellation processing execution unit that executes call setup / cancellation processing in the processing units 10a and 10b, 16 is a bus interface unit, 17 is an input / output interface unit, and 18 is a processor.
Reference data provided in 1b- ₁ and 1b- ₂ . Also, 21
Reference numerals a and 21b are management processors, and reference numeral 22 is reference data provided in the management processor 21a.

FIG. 2 is a block diagram of an embodiment in which reference data is provided in a management processor (hereinafter referred to as MPR) 21a. In the figure, terminals _4-10 to 4 are shown as a plurality of processors.
_-19 and 4 _-20 to 4 _-29 connected (10 lines and terminals are shown only partially) 3 _{-10 to} 3 _-19 , 3
Two processors 1 each containing _{-20 to} 3 _-29
Only b _-1 , 1b _-2 are shown.

Hereinafter, processor 1a _-1 (hereinafter CPR # 1
And referred) to stowed terminal 4 _-10 (hereinafter, terminal T referred to as ₁₀₎ by the processor 1b _-2 (hereinafter, CPR # 2 and referred) to accommodated the terminal 4 _-20 (hereinafter referred to as the terminal T ₂₀ ) Is called and packet communication is performed between both terminals, the description will be simplified for the same parts as those in the prior art.

The congestion state management unit 11 in CPR # 1 and CPR # 2 in FIG. 2 realizes the congestion state management means in FIG. 1, and periodically monitors the processing state of the processing unit 10a, It stores whether or not it is in a congestion state. Whether or not the congestion state is determined by a known method such as monitoring the idle time of the processing unit 15a, for example, when the idle time is a predetermined time or more, the congestion state of its own processor in the congestion state storage table 16 Information indicating a non-congestion state (for example, "0") is stored in a storage column (not shown), and information indicating a congestion state (for example, "1") when the idle time is less than a predetermined time. Is recorded as congestion state information.

When the change from the non-congestion state to the congestion state is confirmed in the regular monitoring, the congestion state information transmission / reception unit 15 sends the congestion state information to all other processors via the bus interface unit 16. It is sent to notify that the congestion status of the local processor has changed.

For example, when congestion state information is sent from the CPR # 2 to the bus 2, the congestion state transmission / reception unit 15 of the CPR # 1
Receives this congestion state information from the bus interface unit 16 and stores it in the congestion state information storage column of CPR # 2 of the congestion state storage table 14. In this way, the congestion state storage table 14 of each processor always stores the congestion states of its own processor and all other processors.

In the above state, the terminal T ₁₀ is connected to the terminal T _20.
When a CR packet is sent out for packet communication with, the call setting processor selection unit 12 of CPR # 1 receives the CR packet via the input / output interface unit 17.

Upon receiving the CR packet, the call setting processor selection unit 12 accesses the congestion state storage table 14 and confirms whether or not the own processor (CPR # 1) is in the congestion state. As a result, when it is confirmed that the own processor is not in the congestion state, the processing unit 10 processes the received CR packet.
transfer to a. In the processing unit 10a, a call setup / cancellation processing execution unit
13 receives this CR packet and starts the call setup process. The subsequent operation of the call setup / cancellation processing execution unit 13 is the same as that of the call setup / cancellation processing execution unit 53 shown in FIG. The processor that has performed the call setting process stores information about the call in a call control memory (not shown) as in FIG. 5, and manages the call until the end.

When the call setting processor selection unit 12 accesses the congestion state storage table 14, its own processor (C
If it is confirmed that the PR # 1) is in the congestion state, the call setting processor selection unit 12 causes the congestion state storage table 14
One of the processors whose congestion state information indicates that it is in a non-congested state is selected by a predetermined method. In this case, the processor to be selected is selected irrespective of the processor accommodating the called terminal T _20, but for convenience of explanation below, CPR # 2 is selected as the processor in the non-congested state. As described below.

The call setting processor selection unit 12 uses the CPR #
If you select 2, the received CR packet will
Transfer to CPR # 2 via the interface 16 and
End T _TenCPR # 2 for call setup / cancellation processing for outgoing calls
The request is stored in a memory (not shown). The C
The R packet is sent to the processing unit 10a of CPR # 2 via the bus 2.
Received by the call setup / cancellation processing execution unit 13 in the processing unit 10a.
To be done.

Upon receiving the CR packet, the call setup / cancellation processing execution unit 13 of the CPR # 2 receives the reference data 22 of the MPR 21a.
To the called terminal T shown in the CR packet.
Check which processor the ₂₀ is in.

As a result, the called terminal T ₂₀ has its own processor,
That is, when it is confirmed that it is contained in CPR # 2,
The call setup / cancellation process execution unit 13 confirms via the call setup process processor selection unit 12 whether or not its own processor is in a congestion state. CPR # 2 was requested by the CPR # 1 to execute call setup / cancellation processing because it was confirmed that it was in a non-congested state. However, the congestion state may have changed over time, so again It is something to confirm.

As a result, if it is confirmed that the self processor is not in the congestion state, the call setting processor selection unit 12
Notifies the call setup / cancellation processing execution unit 13 of the confirmation result, and the call setup / cancellation processing execution unit 13 starts the call setup processing. In this case, since the processor not accommodating the calling terminal T ₁₀ performs the call setting process, the process of the call setting / cancelling process executing unit 13 is performed by the call setting / cancelling process executing unit of the MPR 41 shown in FIG. Same as 42, but since the specific operation is almost the same except that the processor is MPR or CPR, detailed description is omitted.

CPR # 2 call setting processor selection unit
When 12 confirms that its own processor is in a congested state, the requesting CPR is sent via the bus interface unit 16.
Return the processing request to # 1. When the call setting processing processor selection unit 12 of the CPR # 1 receives this via the bus interface unit 16 of its own processor, it accesses the congestion state storage table 14 and selects another processor that is not in the congestion state, and the bus interface The third processor (not shown) selected via the unit 16 is requested to execute the call setup / cancellation process again.

In the following, description will be given assuming that CPR # 2 is not in the congestion state. In this case, the call setup / cancellation processing execution unit 13 of CPR # 2 performs the call setup processing, and the data necessary for the call setup / cancellation processing of the packet communication between the terminals T ₁₀ and T ₂₀ is omitted from the control memory. Then, the call setting process ends. As a result, data packet transmission / reception is started between the terminals T ₁₀ and T ₂₀ .

Packet communication between the originating and terminating terminals T ₁₀ and T ₂₀ is performed by the flow control processing units (not shown) of CPR # 1 and CPR # 2. In this processing, CPR # 1 is used.
Storing the call corresponding memory is not shown the transmission packet amount (or the received packet amount from the terminal T ₁₀₎ relative to the CPR # 2, CPR # 2 is also self-transmission packet amount with respect to the terminal T ₂₀ It is stored in a call corresponding memory (not shown) in the processor.

The call release process is performed by the call setup / release process execution unit 13 of the processor that has executed the call setup process. Therefore, when packet communication is completed and a CQ packet is sent from the terminal T ₁₀ to CPR # 1, the CPR The call setting processor selection means 12 of # 1 is sent from the bus interface unit 16 to the CP.
Transfer the CQ packet to R # 2.

Upon receiving the CQ packet via the bus interface unit 16, the call setup / cancellation process execution unit 13 of CPR # 2 starts the call cancellation process. Call setup / release processing execution unit
The call release process by 13 is shown in FIG.
This is similar to the call release processing by the call setup / cancellation processing execution unit 42 of the MPR 41, and since the specific operation is almost the same except that the processor is MPR or CPR, detailed description is omitted. To do.

FIG. 3 shows the configuration of an embodiment in which the reference data necessary for call setup / cancellation processing is not provided in common to the management processor (MPR) but is distributed to each processor (CPR). In this configuration, each of the processors 1b- ₁ to ₁ has reference data 18 therein.

Therefore, for example, the call setting process when the terminal T ₁₀ makes a call is executed by the call setting / cancelling process execution unit 13 of CPR # 2.
Call setup / cancellation processing execution unit of CPR # 2
When the CR packet from the terminal T ₁₀ is transferred from the CPR # 1, 13 accesses the reference data 18 in its own processor without accessing the MPR 21b, and accommodates the called terminal T ₂₀ indicated in the CR packet. Check the processor to be used. The operation of FIG. 3 is the same as that of FIG. 2 except for this point, and the other description will be omitted.

As described above, in the configurations of FIGS. 2 and 3, the plurality of processors 1a _-1 to 1b _-1 all include the call setup / cancellation processing execution unit 13, but the plurality of terminals 4 _-10. When a call is made from any of the above, the processor accommodating the calling terminal does not always execute the call setup / cancellation process, and the processor not in the congestion state executes the call setup / cancellation process.

Therefore, even when calls are concentrated from a specific terminal such as a host computer, only the processor accommodating the specific terminal does not become congested, and the load is distributed to a plurality of processors.

Although the embodiment of the present invention has been described above with reference to FIGS. 2 and 3, it goes without saying that the present invention is not limited to the illustrated construction or the operation described above.

For example, in the above description, it is described that the congestion state storage table 14 stores "1" or "0" as to whether or not each of the processors 1a- ₁ to 1b- ₁ is in the congestion state. Stores the congestion state of each processor 1a _-1 ~, 1b _-1 ~ with a number indicating the processing efficiency of the processing unit 10a or 10b,
It is easy to realize that the call setting processor selection unit 12 selects the processor with the lowest processing efficiency by a known technique. When such a configuration is adopted, it is possible to further average the load distribution among the plurality of processors 1a- ₁ to 1b- ₁ and the effect of the present invention becomes more remarkable.

In the above description, the billing process is performed in the call release process, but the billing process may not be necessary depending on the usage of the packet switch. However, it is clear that the effect of the present invention does not change depending on whether or not the charging process is performed.

Further, in the above, as a result of the processor (for example, CPR # 2) requested to execute the call setup / cancellation processing being in the non-congested state reconfirming the congestion state of its own processor, it is in the congestion state. If it is confirmed, the processing was returned to the requesting processor, but without returning to the requesting processor, a third processor that is not in the congestion state in CPR # 2 is selected and the call setup / release processing is executed. The effect of the present invention does not change even if the processor requested to execute the call setup / cancellation process does not reconfirm the congestion state of its own processor.

Although the management processor 21b is provided in FIG. 3, unlike the management processor 21a in FIG. 2, it is not necessary to provide reference data in the management processor 21b. Therefore, in the case of FIG. Even if omitted, the effect of the present invention does not change.

Further, in FIG. 3, all the processors 1b- ₁ to
Although reference data 23 provided as described on the assumption that it has a reference data to all common, and only the data common to routing such as whole processors required to call setup / release process and the common terminal 4 _- It is also possible to limit the data regarding the service characteristics of ₁₀ to only the data of the terminal accommodated in the own processor. In this case, each time the call setup / cancellation processing execution unit 13 that executes the call setup / cancellation processing needs to check the data of the required terminal by accessing the processor accommodating the terminal,
Even in this case, the effect of the present invention does not change.

[0067]

As described above, according to the present invention,
In a multiprocessor type packet switch configured by connecting a plurality of processors each accommodating one or a plurality of lines via a bus, a large number of calls are made from a plurality of lines or a call is made from a specific line. Even if it is done in a concentrated manner, the load will not be concentrated on a specific processor, so the loads on multiple processors will be averaged, and the congestion of a specific processor will not limit the processing capacity of the packet switch. Therefore, the processing capability of the packet switch is significantly improved.

[Brief description of drawings]

FIG. 1 is an explanatory view of the principle of the present invention.

FIG. 2 is a configuration diagram of an embodiment of the present invention (No. 1)

FIG. 3 is a configuration diagram of an embodiment of the present invention (No. 2)

FIG. 4 is a block diagram of a conventional technique (No. 1)

FIG. 5 is a configuration diagram of a conventional technique (No. 2)

[Explanation of symbols]

1 _-1 to 1 _-n processor 2 bus 3 line 11 congestion state management means 12 call setting processor selecting means 13 call setting / cancelling processing executing means

Claims (1)

[Claims] 1. A plurality of processors (1 _-1 to 1) each accommodating one or a plurality of lines (3) and processing packet communication.
_-n ) is a processor load balancing method in a multiprocessor packet switch in which a bus (2) is connected to each of the plurality of processors (1 _-1 to 1 _-n ) Congestion state management means (11) that monitors the state, stores it as congestion state information and sends the congestion state information to another processor, and receives and stores the congestion state information sent from another processor, and When there is a call from the line (3), the congestion status management means (11) checks the congestion status of each processor (1 _-1 to 1 _-n ), and if the processor is not in the congestion status, Call setup / cancellation processing is requested to the processor call setup / cancellation processing execution means (13), and when the processor itself is in a congested state, one of the processors not in the congested state is selected and the call setup / cancellation processing is performed. Call to request execution of The fixed processing processor selection means (12) and the call setting processing execution processor selection means (12) request the execution of the call setting / cancellation processing, or another processor requests the execution of the call setting / cancellation processing. Call setup / cancellation processing execution means for executing call setup / cancellation processing at times (13)
And a load distribution method for a packet switch, characterized in that the load of a processor associated with call setup / cancellation processing is distributed to a plurality of processors.