JP2000148525A - Method for reducing load of active system in service processor duplex system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent processing performance from being deteriorated at the time of high load by reducing the load of an active system service processor(SVP) in a hot stand-by type duplex SVP. SOLUTION: Instruction-sorted performance request discriminating tables 114, 124 respectively defining devices taking charge of the processing of an active system and a stand-by system in each instruction are included in respective SVPs 110, 120. When a processing request is issued from a computer system (CPU) 100, the active SVP (defined as 110) refers to the table 114, and when the device taking charge of processing is the active system, it executes the request and reports the processing result to the CPU 100, but when the device taking charge of processing is a stand-by system, it requests the stand-by SVP (defined as 120) to execute the processing of the request. The stand-by SVP 120 executes process-ing requested from the active SVP 110 and reports its processing result to the active SVP 110, which outputs the received process-ing result to the CPU 100.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a redundant service processor system, and more particularly to a method for reducing a load on an active system in a redundant service processor system in a hot standby system.

[0002]

2. Description of the Related Art A conventional processing method in a system in which service processors for performing services such as power control and failure monitoring of a computer system are duplicated by a hot standby method will be described with reference to FIG. In FIG. 5, a computer system (CPU) 100 includes a service processor 1
10, 120 are connected in a potato style. The service processors 110 and 120 are separately connected by a service processor interface line 130. Now
The service processor 110 is an active system, and the service processor 120 is a standby system.

Usually, the active service processor 110
Processes the request (instruction) from the computer system 100 (500), and reports the processing result to the computer system 100. One of the standby service processors 120 uses the inter-service processor interface line 130 to perform polling and current operation. The active service processor 120 is monitored by a failure interrupt from the system. When a failure occurs in the active service processor 110 and the failure is detected by the standby service processor 120 (501), the active service processor 110 becomes the standby system by the service processor replacement (502) (503), and the standby service The processor 120 becomes the active system (504), and the service processor 120 processes subsequent requests (commands) from the computer system 100 (505) and reports the processing result to the computer system 100.

[0004]

The hot standby system is a system for eliminating a single point and aiming at high reliability. However, the prior art does not consider the performance and the performance is a single service processor. Average was the limit. That is, since all the processing requests from the computer system are handled by the active service processor, when processing requests from the computer system to the active service processor are concentrated, the processing performance of each processing is lower than usual. there were.

An object of the present invention is to provide a system in which service processors for performing services such as power control and fault monitoring of a computer system are duplicated by a hot standby system.
This reduces the load on the active service processor and prevents the processing performance of the active service processor from decreasing when the load is high.

[0006]

In order to achieve the above object, in the present invention, a part of the processing of the active service processor is processed by the standby service processor. Specifically, processing in the active system and processing in the standby system are defined in advance, and the active service processor
Determines the processing request from the computer system, executes the processing of the active system by its own service processor, reports the processing result to the computer system, and requests the processing of the standby system to the processing processor of the standby system to perform the next processing. The standby service processor executes the process requested by the active service processor and reports the processing result to the active service processor.The active service processor receives the request from the standby service processor. The reported processing result is sent to the computer system.

[0007]

DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of a duplicated service processor system according to the present invention. In the figure, a computer system (CPU)
100, service processors (SVP) 110, 12
0 is connected in the form of a potato. Service processor 1
10 includes a CPU transmission / reception control module 111, a main module 112, a communication control module 113, and a performance requirement determination table 114 for each instruction. The CPU transmission / reception control module 111 has an interrupt mask 111-1. Wait flag 11
2-1. There is a standby state flag 112-2. The configuration of the service processor 120 is as follows.
Is the same as The communication control modules 113 and 123 of both service processors 110 and 120 are connected through an inter-service-processor interface line 130. Further, the service processors 110 and 120 include an external storage device such as a shared hard disk 140. In addition, the service processors 110 and 120 generally include a console device such as a keyboard and a display, but are omitted in FIG. The service processors 110 and 120 may share one console device or the like.

[0008] The interrupt masks 111-1 and 121-1 are:
Indicates permission / prohibition of interruption from the computer system 100,
The active system takes an interrupt enabled state, and the standby system takes an interrupt disabled state. The active / standby flags 112-1, 122-1 are
Indicates whether the own service processor is an active system or a standby system. The current / standby flags 112-1, 122-1
Reflects the state of the interrupt masks 111-1 and 121-1 as they are. Standby state flags 112-2, 122-
Reference numeral 2 indicates whether the standby processor of the partner is normal or faulty when the own service processor is the active system.

[0011] Instruction-specific performance requirement determination tables 114 and 12
Reference numeral 4 denotes a table storing information defining processing in the active system and processing in the standby system. FIG. 2 shows a specific example.
FIG. 2 shows, for each instruction, whether the device in charge of processing is an active service processor or a standby service processor. The information (processing SVP definition information) stored in the instruction-specific performance requirement determination tables 114 and 124 is created together with, for example, predicting the load on the service processor when creating the processing program of the computer system.

First, the service processors 110, 120
Is determined by the operator or the like as the active / standby system, and the hard disk 140 is used as interrupt mask information.
Is stored in Also, the hard disk 140 has
Processing S for instruction-specific performance requirement determination tables 114 and 124
VP definition information is also stored. When starting the system,
The interrupt mask information of the hard disk 140 corresponds to the interrupt masks 111-1 and 111-1 of the service processors 110 and 120.
21-1 which is the active / standby flag 112-
1, 122-1. At the same time, the processing SVP definition information of the hard disk 140 is loaded into the per-instruction performance requirement determination tables 114 and 124. The contents of the instruction-specific performance requirement determination tables 114 and 124 are the same.

First, the overall operation of FIG. 1 will be described. Here, the service processor 110 is an active service processor, and the service processor 120 is a standby service processor. In this case, the active service processor 1
In 10, the interrupt mask 111-1 of the CPU transmission / reception control module 111 indicates a state in which an interrupt from the computer system 100 is permitted, and the active / standby flag 112-1 of the main module 112 indicates that the own service processor 110 is the active system. . On the other hand, in the standby service processor 120, the interrupt mask 121-1 of the CPU transmission / reception control module 121 indicates a state in which the interrupt from the computer system 100 is prohibited, and the active / standby flag 122-1 of the main module 122 indicates that the own service processor 120 is in the standby state. Indicates a system.

The computer system 100 requests the service processors 110 and 120 to perform processing by interruption. The CPU transmission / reception control module 111 of the active service processor 110 determines that the interrupt mask 111-1 indicates that the interrupt is permitted.
, And passes a processing request to the main module 112. On the other hand, the standby service processor 120
The CPU transmission / reception control module 121 prohibits the interruption from the computer system 100 and discards the processing request because the interruption mask 121-1 indicates the interruption prohibition state.

The main module 112 of the active service processor 110 includes a CPU transmission / reception control module 11
When a processing request (instruction) is passed from 1, the request is processed by referring to the per-instruction performance requirement determination table 114 and if the request is in charge of the active system, and the processing result is passed through the CPU transmission / reception control module 111. The request is reported to the computer system 100. If the request is assigned to the standby system, the request refers to the standby state flag 112-2. If the standby service processor 120 of the other party is normal, the request is transmitted to the communication control module 113. The data is transmitted to the standby service processor 120 through the service processor interface line 130. When receiving the processing request from the active service processor 110 through the communication control module 123, the main module 122 of the standby service processor 120 processes the request and transmits the processing result to the communication control module 1
23. Report to the active service processor 110 through the service processor interface line 130 (processing end report). Transmission and reception of information between the communication control modules 113 and 123 are performed, for example, in a packet format. The main module 112 of the active service processor 110 includes:
When the processing result from the standby service processor 120 is received through the communication control module 113, it is transmitted to the CP.
A report is sent to the computer system 100 through the U transmission / reception control module 111. If the standby service processor 1
If a failure occurs during the processing requested by the active service processor 110, the communication control module 113 of the active service processor 110 monitors the timeout for each processing, and the processing result is not returned within a certain time. If not, a processing failure is reported to the main module 112. This is called the main module 112
Performs processing in its own system when a processing failure due to timeout is reported.

On the other hand, the communication control module 123 of the standby service processor 120 issues polling to the partner communication control module 113 through the service processor interface line 130 at regular intervals to monitor the active service processor 110. When a failure of the active service processor 110 is detected, the failure is reported to the main module 122. When a failure occurs in the active service processor 110, the active service processor 110 passes through the service processor interface line 130.
Raises a fault interrupt to the standby service processor 120,
This failure interrupt is detected by the communication control module 123 of the standby service processor 120 and is reported to the main module 112. Upon receiving the failure report of the active service processor 110 from the communication control module 123, the main module 122 changes the interrupt mask 121-1 of the own CPU transmission / reception control module 121 to the interrupt enabled state, and the active / standby state of the own main module 122 is changed. The flag 122-1 is set to the active system, and the standby state flag 122 is set.
-2 is assumed to be faulty. In the active service processor 110 in which a failure has occurred, the main module 112 autonomously or the standby service processor 120 forcibly restarts the active service processor 110 so that the interrupt mask of the CPU transmission / reception control module 111 is interrupted. 111-1 is in the interrupt disabled state, and the active / standby flag 112-1 of the main module 112 is set to the standby system. As a result, subsequent requests from the computer system 100 are accepted by the new active service processor 120.

Thereafter, when the failure of the service processor 110 is recovered, it is optional whether the service processor 120 is continuously used or the service processor 110 is returned to the active system.

FIG. 3 shows a service processor (SVP) 1
It shows an example of a processing flow particularly related to the present invention in the main modules 112 and 122 of the modules 10 and 120. Here, the case of the main module 112 of the service processor 110 will be described.

When the main module 112 receives any processing request (step 301), it determines whether or not its own device is an active service processor (step 30).
2). If the own device is the active service processor, it is determined whether or not the processing is a process completion report from the standby service processor through the communication control module 113 (step 30).
3). In the case of a processing end report from the partner standby service processor, the processing result of the standby service processor is
A report is sent to the computer system 100 through the CPU transmission / reception control module 111 (step 304), and
Return to 1.

If it is not a processing end report from the standby service processor, that is, the CPU transmission / reception control module 1
In the case of a processing request from the computer system 100 through No. 11, referring to the instruction-specific performance request determination table 114,
It is determined whether the device in charge of processing the request (command) is an active system or a standby system (step 305). If it is an active system, the request is processed as it is (step 306), and the processing result is
A report is sent to the computer system 100 through the U transmission / reception control module 111 (step 304), and the process returns to step 301. On the other hand, if the device in charge of processing is the standby system, it is determined whether the partner standby system service processor is normal (step 307). If not (in the event of a failure), the request is processed as it is, and the processing result is obtained. Computer system 1
00 (step 304), and returns to step 301. When it is confirmed that the partner standby service processor is normal, the request is sent to the partner standby service processor through the communication control module 113 to request processing (step 308), and the process returns to step 301.

On the other hand, if the own device is a standby service processor, the main module 112 executes the received processing (step 309). Then, it is determined whether the processing request is a request from the active service processor (step 310). If the processing request is a processing request from the active service processor, the processing result is transmitted through the communication control module 113 to the other active service processor. (Step 311), and returns to step 301. If the request is not a processing request from the active service processor (for example, a maintenance request by an operator), after executing the processing, the process skips step 310 and returns to step 301.

FIG. 4 is a table showing the performance requirement discrimination table 11 for each instruction.
4 and 124 by applying the contents of FIG.
FIG. 9 is a diagram showing a specific processing flow in each service processor 110 and 120 when a processing request (instruction) is issued from 00. In FIG. 4, the CPU transmission / reception control modules 111 and 121 are omitted.

Hereinafter, how the service processors 110 and 120 process instructions from the computer system 100 to realize the present invention will be described with reference to FIG. Also in this case, the service processor 110 is an active service processor, and the service processor 120 is a standby service processor.

When the instruction 1 is issued from the computer system 100 (400), the main module 112 of the active service processor 110 changes from a processing reception waiting (410) to a processing reception (411) and starts processing. First, the main module 112 refers to the per-instruction performance requirement determination table 114, judges that the instruction 1 is to be processed by the standby service processor 120 (412), and confirms that the other standby service processor 120 is normal. Then, the processing request to the standby service processor 120 is notified to the communication control module 113 (413), and the process returns to the processing reception waiting (414). The communication control module 113 that has received the notification sets the processing request content in the communication packet,
The data is sent to the standby service processor (430).

The communication control module 123 of the standby service processor 120 includes the active service processor 11
When the communication packet from 0 is received, the main module 122 is notified (440). The main module 122 of the standby service processor 120 changes from the processing reception waiting (450) to the processing reception (451), executes the requested processing (452), and sends a result report request to the active service processor 110 to the communication control module 123. (453), and returns to waiting for processing (454). The communication control module 123 that has received the notification sets the processing result in the communication packet and transmits it to the active service processor (441).

The communication control module 113 of the active service processor 110
When the communication packet from 0 is received, it notifies the main module (431). Active service processor 11
0, the main module 112 waits for processing to be accepted (41
From 9), the process is accepted (421), and if the content of the process request is determined to be the process result from the standby service processor 120, the process result is reported (421) to the computer system through the CPU transmission / reception control module 111, and the process is waited for (422). ).

On the other hand, when the computer system 100 issues the instruction 1 after issuing the instruction 1 (401), the main module 112 of the active service processor 110
Are waiting for processing (414) and then receiving processing (41).
5), and the process starts. Main module 112
Determines with reference to the instruction-specific performance requirement determination table 114 that the instruction 2 is to be processed by the active service processor 110 (416), and executes the processing as it is (417).
The processing result is reported to the computer system 100 via the CPU transmission / reception control module 111 (418), and the process returns to waiting for processing (419).

As described above, the instruction 1 and the instruction 2 are separated from each other by using the hot standby system.
By performing the processing in steps 10 and 120, the load on the active service processor can be reduced.

[0027]

According to the present invention, in a system in which service processors of a computer system are duplicated by a hot standby system, processing of the active service processor under a heavy load while ensuring high reliability by the hot standby system. It is possible to prevent the performance from dropping, and the performance can be improved more than before.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of a redundant service processor system according to the present invention.

FIG. 2 is a diagram illustrating a specific example of a performance requirement determination table for each instruction.

FIG. 3 is an example of a processing flow of a main module common to an active service processor and a standby service processor.

FIG. 4 is a diagram showing a specific processing flow in each service processor in response to a processing request from a computer system.

FIG. 5 is a diagram illustrating a process takeover at the time of occurrence of a failure by the hot standby method.

[Explanation of symbols]

 REFERENCE SIGNS LIST 100 computer system 110, 120 service processor 111, 121 CPU transmission / reception control module 112, 122 main module 113, 123 communication control module 114, 124 per-instruction performance requirement determination table 130 interface line between service processors

Claims (1)

[Claims] In a system in which a service processor for executing various service processes of a computer system is duplicated between an active system and a standby system, processing in the active system and processing in the standby system are defined in advance, and the active system is defined in the active system. The service processor determines the processing request from the computer system, executes the processing of the active system by its own service processor, reports the processing result to the computer system, and requests the processing of the standby system to the processing system of the standby system. Reporting the processing result of the standby service processor to the computer system, executing the processing requested by the active service processor, and reporting the processing result to the active service processor; A method for reducing the load on an active system of a redundant service processor system.