System for preserving logical partitions of distributed parallel processing system after re-booting by mapping nodes to their respective sub-environments

• US 5,854,896 A
• Filed: 06/17/1996
• Issued: 12/29/1998
• Est. Priority Date: 06/17/1996
• Status: Expired due to Term

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Brenner, Larry Bert, Rothaupt, Krystal Kay, Briskey, Kenneth Charles
• Primary Examiner(s): Meky, Moustafa M.
• Application Number:
  US08/664,577
• Time in Patent Office: 925 Days
• Field of Search: 395/200.5, 395/200.51, 395/200.53, 395/200.54, 395/200.52, 370/241, 370/245, 370/252, 370/254
• US Class Current: 709/221
• CPC Class Codes: G06F 9/4416 Network booting; Remote initial program loading [RIPL]

Abstract

In accordance with a preferred embodiment of the present invention, an application and method of preserving persistent logical partitions, is suggested for use in a partitioned massively parallel processing system environment. All system data, including configuration data is stored in a central control element in a special configuration object class file. The configuration object class file maps the resources to different sub-environments. Upon a system shut-down or re-booting of any resource, the configuration information stored in the central control element is recovered to re-establish the sub-environments integrity.

Claims

1. In a distributed parallel processing system environment having a plurality of nodes connected to one another via a network in a manner as to form a single system unit initially, said environment being then logically partitioned into at least two sub-environments, a method of preserving said logical partitions after re-booting, said method comprising the steps of:

controlling and maintaining all communications and status information between nodes through a central control element connected to said nodes via said network;

storing all system data in a system data repository located in said central control element;

using said stored data, creating a special configuration object class file in said system data repository, said object class file mapping said nodes to their respective sub-environments; and

retrieving any portion of said system data relating to configuration and providing it to each node after rebooting completion.

2. The method of claim 1, wherein said configuration object class file is defined as a Syspar_-- map class, accessible to all nodes in all said sub-environments.

3. The method of claim 1, wherein said system environment is partitioned into a plurality of sub-environments.

4. The method of claim 1, wherein said central control element comprises a communication adapter.

5. The method of claim 4, wherein said communication adapter is apportioned in a manner so that information about each of said sub-environments is mapped in different locations of said adapter, said portions being identified by their respective addresses.

6. The method of claim 5, wherein said adapter portions are further identified by a host-name.

7. The method of claim 6, wherein each node is provided a destination file, said destination file having host-name of said adapter portion assigned to its respective sub-environment.

8. The method of claim 5, wherein each node is provided a destination file, said destination file having address of the adapter portion assigned to its respective sub-environment.

9. The method of claim 8, wherein said central control element can determine said node'"'"'s sub-environment just by examining address of adapter portion in said node'"'"'s destination file.

10. The method of claim 9, wherein said central control element examines said address in said destination file and compares it to configuration information stored in said configuration object class file after re-boot completion.

11. The method of claim 1, wherein said network is an ethernet using a TCP/IP protocol.

12. In a distributed parallel processing environment having a plurality of nodes connected to one another via a network in such way as to form a single system unit, said environment logically partitioned into at least two sub-environments, a method of restoring a node to its respective sub-environment after a node re-boot, said method comprising the steps of:

controlling and maintaining all communications and status information between said nodes through a central control element connected to said nodes via said network;

storing all system data in a system data repository located in said central control element;

using said stored data, creating a special configuration object class file in said system repository, said configuration object class file mapping said nodes to their respective sub-environments; and

retrieving any portion of said system data relating to configuration information and providing it to any node after re-booting completion.

13. The method of claim 12, wherein said node is provided all other necessary information belonging to its sub-environment alongside said configuration information.

14. The method of claim 12, wherein a plurality of nodes have simultaneously completed re-booting, and configuration information are provided to them about their respective sub-environments.

15. In a distributed parallel processing environment having a plurality of resources connected to one another via a network in such way as to form a single system unit, said environment logically partitioned into at least two sub-environments, a method of preserving said logical partitions after re-booting said system environment, said method comprising the steps of:

controlling and maintaining all communications and status information between said resources through a central control element connected to said nodes via said network;

storing all system data in a system data repository located in said central control element;

using said stored data, creating a special configuration object class file in said system repository, said configuration object class file mapping said resources to their respective sub-environments; and

retrieving an portions of said system data dealing with configuration information and providing it to each resource after re-booting completion.

16. The method of claim 15, wherein said resources are a plurality of nodes and a plurality of switches in direct processing communications with one another.

17. A distributed parallel processing system environment having a plurality of nodes connected to one another via a network in such way as to form a single system unit, said environment further partitioned into at least two logically persistent sub-environments comprising:

a central control element for controlling and maintaining all communications and status information between said nodes, said central control element connected to said nodes via said network;

a configuration data repository for storing all configuration data located in said central control element; and

a system object class file created in said repository for storing configuration data, said system object class mapping all said nodes to their respective sub-environments.

18. The system environment of claim 17, wherein said central control element comprises a communication adapter.

19. The system environment of claim 17, wherein said communication adapter is apportioned in a manner so that information about each of said sub-environments is mapped in different locations, identified by dedicated addresses.

20. The system environment of claim 19, wherein said portions in said communication adapter are further defined by host-names.

21. The system environment of claim 19, wherein each node is provided a destination file, said destination file having address of the portion of said communication adapter assigned to its respective sub-environment.

22. The system environment of claim 17, wherein said system environment is partitioned into a plurality of sub-environments.

23. The method of claim 17, wherein said network is an ethernet using a TCP/IP protocol.

24. The system environment of claim 17, wherein each said sub-environments also comprises of at least one dedicated switch, said switches in processing communication to said nodes directly, and to each other via said network.

25. The system environment of claim 24, wherein any configuration data about said switch is also provided in said configuration object class file and restored upon a re-boot completion.