JPH03209533A - Redundant system consisting of plural processors - Google Patents

Abstract

PURPOSE:To obtain a redundant system having high reliability without duplexing totally the system by providing the processors into the system as the redundant components together with a cooling device, etc. CONSTITUTION:If one of processors(PRC0 - PRCn) 110 - 11n has a fault in a system operating state, a system unit controller SUC 5 applies an interruption to a host OS to separate the faulty processor from other nondefective processors. Then the SUC 5 cuts off the supply of power to the faulty processor and separates this processor from a system. At the same time, the SUC 5 orders the optional one of nondefective processors PRC0 - n to load a microprogram via a microcode storage MSC 7 as long as the system operation can be carried on with reduction of the system operating scale. So that a spare processor (PRCn + 1) 11n + 1 or the optional nondefective processor is started. In such a constitution, the reliability of a redundant system is improved.

Description

[Detailed Description of the Invention] [Summary] Conventionally, regarding a stem having a plurality of microprogram-controlled processing devices and a cooling device for each of the processing devices, if any one of the processing devices or the cooling device breaks down, In order to solve this problem, measures such as duplicating the entire system were installed, and one or more spare processing units and cooling units were installed, and when the system was started up, the system was stopped. means for storing system configuration information to be sent; and means for backing up initial microprogram code to be sent to each processing unit;
/If a failure occurs during system operation, the above-mentioned spare processing device and cooling device are used as alternative devices, and the above-mentioned backed-up microprogram code is loaded into the alternative processing device and started up. Configure.

[Industrial Field of Application] The present invention relates to a super-large computer system or a system consisting of a plurality of microprogram-controlled processing devices such as a supercomputer, and in particular, each processing device has a cooling device, and furthermore, the above-mentioned processing The present invention relates to a redundant system consisting of a plurality of processing devices having one or more spare devices for each device and cooling device.

[Prior Art] Fig. 2 is a diagram showing an example of the configuration of a conventional system consisting of multiple processing devices, where I is the main storage, 1a is the main storage, and the host OS 12 is a service processor (SVP).
), 3 is a floppy disk (FPD) in which initial microprograms etc. are stored, 4 is a console display device (C5L-DSP), 9 b is a collective refrigerant supply unit (CC5U), 8 e is a control module (CT
M) lOo~IO,, is the refrigerant supply module (C3M)
, 11o-11,, is a processing device (PRC, -PRC,,
), 12, ~12. refers to refrigerant passages such as piping and hoses (CP
.

~CP, , ), 30 is an interface, and 31 is a power control manager + - (Power Contr.

Manager: P CM), 32o ~ 32+,
represents a power supply control device (PC, ~PC,,).

This system consists of n processing units (PRCo to PRC
,,) 110-11. A system having each processing device (PRCo=PRC,,) llo ~11,
, is connected to the host 03Ia via the interface 30, and forms a cooperative system under the management of the host 051a.

In addition, each of the above processing devices (PRC, -PRC,,)
-11, are microprogram-controlled processing units, into which an initial microprogram is loaded when the system is initially started up, and each unit becomes a processing unit for the assigned function.

In addition, each processing device (PRCo to PRCh) 11° to 1
1. A cooling device is provided for each of the 12.
-12. , each processing unit (PRC.

~PRC,,>11°~11. ．． Refrigerant supply module (C3Mo-C5M,...) 1
0 to 10h is used.

In addition, the above refrigerant supply modules (C3MO to CSM,
,) 10o ~10. is the collective refrigerant supply unit (CC
Control module (CTM) 8
Managed by e.

In addition, the power supply control device (PC, -PC,,) 32° ~ 32
．． is each processing device (P RC, ~ P RC,,) ll.

~11. This is a device that controls the power supply to each refrigerant supply module (CsMo-C3M
,,) 10o~10. , to each processing unit (PRC, ~
PRC,) 11. ~11. After confirming that the cooling medium is being returned normally, the power is turned on according to instructions from the power control manager (PCM) 31.

The service processor (SVP) 2 is a processor that operates, manages, and services the system.In addition to loading the initial microprogram when the system is started up, the service processor (SVP) 2 also monitors the operating status of the system and displays the operating status on the console. It is displayed on the display (C5L-DSP) 4 to notify the operator.

[Problems to be Solved by the Invention] In the conventional/stem configuration shown in FIG. 2, for example, the refrigerant supply module
If a failure occurs in any of the refrigerant supply modules, the entire system will stop (the same applies in the case of a failure in the processing device).

In general, systems such as ultra-large computers that are made up of a plurality of processing units often perform important tasks, and a system stoppage would lead to a serious situation.

Therefore, in order to avoid system stoppage, the refrigerant supply module (CSMQ-CSM, ) to.

-10, and refrigerant passage ccpo-CP,,)12°~
In addition to 12 hours, we have taken measures such as duplicating the entire system.

However, duplicating the entire system increases the installation area due to increased hardware, increases costs,
This often leads to complicated control and various other problems.

The present invention has been made in view of the above-mentioned problems, and is a redundant system consisting of a plurality of processing units that can efficiently cope with a failure in some of the system components without duplicating the system. The purpose is to provide a system.

[Means for Solving the Problems] According to the present invention, the above objects are achieved by the means described in the claims.

That is, the present invention provides a method for storing host) O3 stored on the main memory.
A system consisting of n (n≧2) microprogram-controlled processing devices that operate under the management of In a system consisting of a plurality of processing devices that have a service processor for operational management, and are configured to perform an initial microprogram load to each processing device and specify the operational functions of each processing device when the system is started up, One or more spare processing units and cooling units are installed in the system, and the system configuration sent from the service processor at system start-up is sent to an intermediate control unit interposed between the service processor and each processing unit. a system configuration table that stores information;
A means for backing up the microprogram code stored in each processing unit at system start-up, and a means for interrupting the host (host) O3 in the event of a failure of any processing unit or cooling unit during system operation. With permission, based on the information in the system configuration table above,
means for selecting the spare processing equipment and cooling equipment;
This is a redundant system consisting of a plurality of processing devices, in which the backup processing device is provided with means for storing backed-up microprogram codes in an intermediate control device and for starting up the processing devices.

[Function] In the present invention, for a plurality of processing devices and a cooling device corresponding to each processing device, a backup processing device is provided as a substitute when any one of the processing devices or cooling devices breaks down. In addition, an intermediate control device installed between the service processor and each of the above-mentioned processing devices stores the system configuration information sent from the service processor at system startup and the microcontroller sent to each processing device. Memorize the program code.

If any processing device or cooling device fails during system operation, the intermediate control device disconnects the failed device and performs system configuration (Recoding) according to the system configuration information.
For system degradation or system degradation, necessary initial microprogram codes backed up in advance are loaded into the backup processing device and started up.

By adopting such a system configuration, even if a failure occurs during operation, the system can be quickly restored without the intervention of a service processor or the like, and without starting up the system from the beginning.

[Embodiment] FIG. 1 is a diagram showing the /stem configuration of an embodiment of the present invention, and 5 is a system unit controller (System
m lIn+t (:ontroller: S U
C), 6/Describe the stem configuration information 1. 7 is a microcode storage (!) that backs up the initial microprogram code sent from the service processor to each processing unit.
Jicro (:ode Storage: MC
S ), 8a to 8d are control module (CTM, SC
TM, ), 9a is the collective refrigerant supply unit) (CC3U)
, lO,, +1 is the reserve refrigerant supply module (C
8M,,4+),11. , +1 is a backup processing unit (P RC, , ), 12. ,, is the refrigerant passage (C
P,,,),] 3. ~13□1 is the unit controller (Unit Controller: UC)
, 14 a and 14 b are redundant tolerant control devices (TLco, TLC,), 21 is a system configuration information loading bus (Path), 22 is a start/stop signal bus for the collective refrigerant supply unit, and 23 is a refrigerant Supply module CS M.

〜CSM,..., start/stop signal bus, 24 is the power on/off signal bus for the processing device for the unit controllers UC, -UCI,..., 25 is the processing device PR
26 is a microprogram code loading bus for C,-PRC, , , 27 is a microprogram code backup bus (BackυρPath);
represents the OS interrupt bus, and the other symbols are the same as in FIG.

In the system configuration of this embodiment, compared to the conventional system configuration shown in FIG.
,,. , Ml. , and refrigerant supply module (C8M
o, I) lo,,,l are provided, and a processing device (PRCQ~
PRC, ) 11° ~ 11. .

and refrigerant supply module (C3M, -C3M,.

10, ~10. ．． Even if any one of them fails, the system is configured so that it can be switched to the backup device and continue to operate without any trouble.

In addition, the interface and power control manager (PCM) in the conventional system have been abolished, and a new system unit controller (SUC) 5 has been installed.
PRCo-PRC,,)Ilo~11. Services that were previously performed through the interface are now performed through the system unit controller (SUC) 5.

This system unit controller (SUC) 5 has conventional interface functions and power control functions of the power control manager (PCM), as well as system power control, environmental cooling control, and processing units (PRCo to PRC).
,,,) IL ~111. service functions, etc.
This is a major transfer from Service Processor (SVP) 2.

Furthermore, processing bag fit (PRCG -PRC,,,'
) 11o~11°. 1 to the host OS.

Additionally, the system unit controller (StJC) 5 has a system configuration table (CFT) 6 and a microcode storage (MC5) 7 as a backup buffer for microprogram codes, and the system configuration table (CFT) 6 is used for processing units in the system. (PRCO
~PRC,,)llo ~11. Or refrigerant supply module (CS tvL+ -c S Mh) 100-1
0h If any one of them fails, a backup processing unit (P RC,, +) 111. -1 and refrigerant supply module (CSM,, +) 10. ,=1
System Reconfiguration (System Reconfiguration)
ration) or if there is a failure of multiple or more processing units that cannot be covered by backup equipment, the failure of multiple or more processing units (P
RC, -PRC7)llo ~11h or refrigerant supply module (CS M, -CS M,,) 10o-
10. System degradation (Syst.
The microcode storage <MC5)7 is the system 1
．． 1γp, when falling, → Novisbu [] Seza (SV
P) 2 to each processing device (PRC, ~PRC,,) Il
, -II, , to back up the microprogram code sent to.

In the /stem configuration of the above-described embodiment, the operation from startup of the /stem to when a failure occurs during operation will be described below.

However, when the system is powered on, the floppy disk
(FDP) 3 and service processor (SVP) 2c!
:, system unit controller (St, JC) 5, unit controller (UC, , ), control module (CTMo, l), and tolerant controller (TLCo, l)') control circuit power is turned on.

(1) First, the system configuration information stored in the floppy disk (FPD) 3 is transferred to the service processor (S
The system unit controller (S
It is stored in the system configuration table (CFT) 6 in the IJC) 5.

(2) Next, the service processor (SVP) 2 activates the collective refrigerant supply unit (CC3U) 9a via the stem unit controller (SUC) 5.

(3) System unit control (J-RA (SU)
C) Depending on the contents of the system configuration table (CFT) 6 in 5, each refrigerant supply module (CS M, -CSM,
, ) +00-10. , is activated.

(4) Next, /stem unit controller (SIC)
5 is a unit controller (uCo ~ UC,) 130
~13h Start up and monitor the environmental cooling, and if it is normal, the processing device (PRC, -PRC,...) 1
10-11. ．． Turn on the power.

(5) After that, service processor (SVP) 2
Each processing bag contains the microprogram code prepared in advance on the floppy disk (FPD) 3 inside! (PRCO
-PRC,,) 100-10° loading.

At the same time as this loading, microcode storage (
Store and back up the above microprogram code in MC5)7.

By the steps (1) to (5) above, /the startup of the stem is completed /and the system enters the operating state.

(6) Afterwards, during system operation, if a failure occurs in the processing units (P RC, -P RC,,) 1lo-11,, the system unit controller (SU
C)5 interrupts the host O5 and causes the host O3 to issue an interrupt requesting a disconnection instruction for the failure processing device.

(7) Next, the system unit controller (SLI)
C>5 turns off the power to the failure processing device, disconnects the failure processing device from the system, and connects the spare processing device (P
RCH++) 11. , -+ Or, if the scale of the system can be reduced and system operation can continue, load the microprogram from the microcode storage (MC3) 7 to any normal processing unit (PRCO-7). Launch.

In addition, the processing equipment (PRC, -7゜n-
l) 1lo-n-n-1 is any processing device (PRC
.

~PRCo) 110~11. ．． It is configured such that it can be changed depending on the contents of the microprogram in any type of processing unit (for example, "channel" or "rcPU") so that it can function as a backup for the microprogram.

[Effects of the Invention] As explained above, the redundant system comprising a plurality of processing devices of the present invention has the following effects.

(1) By providing redundant processing equipment, cooling equipment, etc. in the system, a highly reliable system can be realized without duplicating the entire system.

(2) Even if more than one processing device or cooling device fails, any normal processing device or cooling device can be substituted for the failed processing device or cooling device to cause system degradation. Continuous operation is possible without downtime.

/Sufu l, System configuration table (CFT) in the unit controller (SUC) 1. By providing i and microcode storage (MC3), switching from a failed processing device to a spare processing device can be quickly performed when a failure occurs.

(3)

[Brief explanation of drawings]

FIG. 1 is a diagram showing a system configuration according to an embodiment of the present invention, and FIG. 2 is a diagram showing an example of the configuration of a conventional system comprising a plurality of processing devices. l... Main storage where host OS is stored, 1a... Host 0812... Service processor (SVP), 3... Floppy disk (FPD), 4... Console display device, 5... System unit controller (SUC), 6... System configuration table (CFT), ? ...Microcode storage (MC5), 8a to gd...Control module (CTMo, CTM,), 9a...
Collective refrigerant supply unit (CCSU), 100-10
．． , l -... Refrigerant supply module (CS M
o -c S Ml,,,'), llo ~11,
,,--...processing device (PRC, -PRC,,,,)
, 12, -12. ,,... Refrigerant passage (CP, ~
CP, , , ), 13o ~ 13. ．． ', ・・・・・・
...Unit controller (UC.

Claims (1)

[Claims] n (n≧2) microprogram-controlled processing devices that operate under the management of a host OS stored in main memory, and n cooling devices that cool each treatment device. The system has a service processor for managing the operation of the system, and when the system is started up, it loads an initial microprogram into each processing unit and controls the operating functions of each processing unit. In a system consisting of a plurality of processing units configured to specify,
The system is provided with one or more spare processing units and cooling units, and furthermore, an intermediate control unit interposed between the service processor and each processing unit contains system information sent from the service processor at the time of system startup. A system configuration table that stores configuration information, a means for backing up the microprogram code stored in each processing unit at system startup, and a means to back up the microprogram code stored in each processing unit when the system is started, and in the event of a failure of any processing unit or cooling unit during system operation. means for obtaining permission to interrupt the host OS and selecting the spare processing device and cooling device based on the information in the system configuration table; What is claimed is: 1. A redundant system comprising a plurality of processing devices, comprising: a means for storing microprogram codes backed up by a computer, and for starting up the processing devices.