Distributed computing utilizing virtual memory having a shared paging space

• US 8,041,877 B2
• Filed: 06/09/2008
• Issued: 10/18/2011
• Est. Priority Date: 06/09/2008
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Rosales, Jacob J., Rosas, Morgan J., Nevarez, David, Patwari, Veena
• Primary Examiner(s): Chery; Mardochee
• Application Number:
  US12/135,762
• Publication Number:
  US 20090307435A1
• Time in Patent Office: 1,226 Days
• Field of Search: 711/147, 711/E12.002
• US Class Current: 711/6
• CPC Class Codes: G06F 12/10 Address translation
• CPC Class Codes: G06F 2009/45583 Memory management, e.g. access or allocation
• CPC Class Codes: G06F 9/45558 Hypervisor-specific management and integration aspects

Abstract

A method for distributed computing utilizing virtual memory is disclosed. The method can include identifying a first node to process an application, identifying paging space accessible to the first node, identifying a second node to share paged data with the first node, and transacting the paged data between the first node and the identified paging space. Thus, application processing results from the first node can be stored in paging space and a second node can retrieve the first result from the paging space such that the paging space can be shared between nodes. Other embodiments are also disclosed.

Claims

1. A method comprising:

identifying a first node configured to process an application;

identifying paging space accessible to the first node;

identifying a second node configured to share paged data with the first node;

transacting the paged data between the first node and the identified paging space;

accessing, by the second node, the paged data from the paging space, wherein the paging space is a paging space of a virtual memory associated with at least one of the first node or the second node, and wherein the first node and second node are configured to share data in a same paging space of the virtual memory;

determining if the second node cannot access the identified paging space directly; and

locating an intermediate node having an intermediate node paging space, of a virtual memory associated with the intermediate node, configured to facilitate sharing of the data between the first and second node in response to determining that the second node cannot access the identified paging space directly.

2. The method of claim 1, wherein the intermediate node paging space is linked to the identified paging space via a virtual input output server.

3. The method of claim 1, wherein the application has a workload that is split into tasks and the first node and the second node utilize the paging space to process the split tasks.

4. The method of claim 3, wherein the first and second node are configured to utilize the paging space to store results of the split tasks.

5. The method of claim 1, wherein a hypervisor facilitates transacting the paged data between the first node and the identified paging space.

6. The method of claim 1, wherein the intermediate node is granted access to data in the paging space and access to data in the intermediate node paging space such that the intermediate node copies data from the paging space to the intermediate node paging space.

7. The method of claim 1, further comprising, in response to a determination that the second node cannot access the identified paging space directly:

controlling the first node to page out contents of a virtual real memory (VRM) stack of the first node to the paging space to thereby generate paged out contents; and

moving the paged out contents of the VRM stack from the paging space to the intermediate node paging space.

8. The method of claim 7, further comprising, in response to the determination that the second node cannot access the identified paging space directly:

paging in the paged out contents of the VRM stack from the intermediate node paging space to a destination paging space of the second node.

9. The method of claim 7, wherein the contents of the VRM stack that are paged out comprise only results of a parsed task of the application processed by the first node.

10. The method of claim 1, further comprising:

transferring the data from the intermediate node paging space of the intermediate node to one or more second intermediate node paging spaces of one or more second intermediate nodes prior to providing the data to the second node.

11. The method of claim 1, wherein determining if the second node cannot access the identified paging space directly comprises determining whether a destination virtual input/output server (VIOS) of the second node can directly access a source VIOS of the first node.

12. A system comprising:

a first node configured to process an application;

a paging space partition module configured to allocate page space to the first node; and

a virtual input output server configured to support the first node in processing the application and to grant access to the page space by the first node and a second node, wherein the page space is a page space of a virtual memory of a virtual memory system, and wherein the first node and second node are configured to share data in the page space of the virtual memory, wherein the virtual input output server is configured to;

determine if the second node cannot access the identified paging space directly; and

locate an intermediate node having an intermediate node paging space, of a virtual memory associated with the intermediate node, configured to facilitate sharing of the data between the first and second node in response to determining that the second node cannot access the identified paging space directly.

13. The system of claim 12, wherein the first node produces a result based on a task of the application and the virtual input output server receives the result and sends the result to the page space of the virtual memory utilizing the paging space partition module.

14. The system of claim 12, further comprising a second virtual input output server configured to access the virtual memory system and retrieve the result via a second paging space partition module.

15. The system of claim 14, wherein the second node receives the result from the second virtual input output server, utilizes the result to produce a second result, and sends the second result to the page space.

16. The system of claim 14, further comprising a hypervisor configured to facilitate communications between the first virtual input output server and the second virtual input output server.

17. The system of claim 12, wherein the paging space partition module comprises an asynchronous services interface.

18. The system of claim 12, further comprising a second virtual input output server to configure communications with the second node.

19. The system of claim 12, further comprising a virtual memory manager configured to accept controls from a hypervisor module and to store the contents in the page space.

20. The system of claim 12, further comprising a stack wherein the paging space partition module utilizes the stack to move application related data and results to facilitate distributed computing between the first and second nodes.

21. The system of claim 12, further comprising a pager to facilitate paging information.

22. A computer program product comprising a computer readable storage medium having a computer readable program stored therein, wherein the computer readable program, when executed on a computing device, causes the computing device to:

process, by a first node, an application;

allocate, by a paging space partition module, page space to the first node;

grant, by a virtual input output server, access by the first node and a second node to the page space, wherein the page space is a page space of a virtual memory of a virtual memory system, and wherein the first node and second node share data in the page space of the virtual memory;

determine if the second node cannot access the page space directly; and

locate an intermediate node having an intermediate node paging space, of a virtual memory associated with the intermediate node, to facilitate sharing of the data between the first and second node in response to determining that the second node cannot access the identified page space directly.

23. The computer program product of claim 22, wherein the computer readable program, in response to a determination that the second node cannot access the identified paging space directly, further causes the computing device to:

control the first node to page out contents of a virtual real memory (VRM) stack of the first node to the paging space to thereby generate paged out contents; and

move the paged out contents of the VRM stack from the paging space to the intermediate node paging space.