US 5,692,119 AGranted

Management method for a multiprocessor system

AssigneeHitachi, Ltd.

InventorsYukio Koguchi, Nobuyuki Suzuki, Toyojiro Hayashi

Life of this patent

Estimating the term…

Loading patent…

Abstract

The number of instruction processors to be independently allocated to each supervisor is limited to a level at which the performance of the tightly coupled multiprocessor system is not deteriorated. A plurality of independent tightly coupled multiprocessor systems comprising these supervisors and a plurality of instruction processors are provided to improve their processing capacity. Since the existence of a plurality of supervisors complicates the management of these supervisors, a supervisor for a total managing service to which separate instruction processors are independently allocated is provided. This supervisor for a total managing service executes initialization and termination of the supervisors and monitoring of a fault, to thereby integrally manage the plurality of supervisors and the on-line control programs under their control. The supervisor manages the resources such as the main storage, the instruction processor and the input/output processors to utilize or manage the CPU's through the management terminals. The supervisor for total managing service, or operation supervisor, manages the status or faults of the supervisor and on-line control programs and provides change-over commands for the on-line control programs.

Claims8

Dependent claims

A multiprocessor control method for controlling a multiprocessor system, wherein said multiprocessor system comprises:

a current multiprocessor system, including:
a plurality of first processors on which a current supervisor and at least one online system are executed, said at least one online system being a program for executing transaction processings under control of said current supervisor, and
at least a second processor on which a current management supervisor is executed, said current management supervisor being a program for controlling start operation of said current supervisor and for detecting a fault in said current supervisor;
a stand-by multiprocessor system including:
a plurality of third processors on which a stand-by supervisor and at least one stand-by online system are executed, said at least one stand-by online system being a program for executing transaction processings under control of said stand-by supervisor, and
at least a fourth processor on which a stand-by management supervisor is executed, said stand-by management supervisor being a program for controlling start operation of said stand-by supervisor and for detecting a fault in said stand-by supervisor, and
a transmission path for connecting said current multiprocessor system with said stand-by multiprocessor system,
wherein said multiprocessor control method comprises the steps of:
detecting, in said stand-by management supervisor, a fault in said current management supervisor;
starting, by said stand-by management supervisor, said stand-by supervisor when said stand-by management supervisor detects a fault in said current management supervisor;
and
starting, by said stand-by supervisor, said stand-by online system to execute said transaction processings.

9 more elements

A multiprocessor system comprising:

a current multiprocessor system including:
a plurality of first processors for operating thereon a current supervisor and at least one online system which executes transaction processings under control of said current supervisor, and
at least a second processor for operating thereon a current management supervisor which executes start operation of said current supervisor and detects a fault in said current supervisor,
a stand-by multiprocessor system including:
a plurality of third processors for operating thereon a stand-by supervisor and at least a stand-by online system which executes transaction processings under control of said supervisor, and
at least a fourth processor for operating thereon a stand-by management supervisor which executes start operation of said stand-by supervisor and detects a fault in said stand-by supervisor, and
a transmission path for connecting said current multiprocessor system with said stand-by multiprocessor system.

4 more elements

Dependencies are inferred from the claim text.

Facts

Application: 07/865,330Filed 8 Apr 1992
Priority: JP 3-07757110 Apr 1991
Granted: 25 Nov 1997
Term ends: Not estimatedThe source records don’t support an estimate.

Classification5 codes

G06F 9/5077
Partitioning or combining of resources: logical partitioning of resources; Management or configuration of virtualized resources
G06F 11/2025
Failover techniques: using centralised failover control functionality
G06F 11/2736
Tester hardware, i.e. output processing circuits: using a dedicated service processor for test
G06F 11/202
Using active fault-masking, e.g. by switching out faulty elements or by switching in spare elements: where processing functionality is redundant