Persistent caching for operating a persistent caching system

• US 9,940,240 B2
• Filed: 10/15/2014
• Issued: 04/10/2018
• Est. Priority Date: 10/23/2013
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Metzler, Bernard, Schmatz, Martin L., Stuedi, Patrick, Trivedi, Animesh K.
• Primary Examiner(s): WANG, LIANG CHE A
• Application Number:
  US14/514,412
• Publication Number:
  US 20150113088A1
• Time in Patent Office: 1,273 Days
• Field of Search: 709212, 709213, 709214, 709215, 709216
• CPC Class Codes: G06F 12/0866 for peripheral storage systems, e.g. disk cache
• CPC Class Codes: G06F 2212/224 Disk storage
• CPC Class Codes: G06F 2212/461 Sector or disk block
• CPC Class Codes: G06F 2212/463 File
• CPC Class Codes: Y02D 10/00 Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

A persistent caching system is provided. The persistent caching system includes a storage system having a caching server for storing data, and a client for accessing the data through a network. The caching server is configured to store the data in a number of virtual memory blocks. The virtual memory blocks refer to an associated memory-mapped file in a file system of the caching server. The caching server is configured to export addresses of the virtual memory blocks to the client. The client is configured to access at least some of the virtual memory blocks through RDMA using the exported addresses. The caching server is configured to page virtual memory blocks being accessed by one or more clients through RDMA to and/or from the memory-mapped files associated with the accessed virtual memory blocks.

Claims

1. A persistent caching system, the system comprising:

a client, wherein the client (a) includes a processor and a computer readable storage medium and (b) is configured to;

i) access at least a portion of a data through a network, and ii) access at least a portion of virtual memory blocks through RDMA using one or more exported addresses; and

a caching server of a storage system, wherein the caching server (a) includes a processor, a computer readable storage medium, and the data, and (b) is configured to;

i) store the data in a number of virtual memory blocks, each of the virtual memory blocks referring to an associated memory-mapped file in a file system of the caching server, ii) export addresses of the virtual memory blocks to the client, and iii) page one or more virtual memory blocks being accessed by the client through RDMA to and/or from memory-mapped files associated with the accessed virtual memory blocks,

wherein, after receiving an RDMA read request from the client at a network card of the caching server, the network card is configured to;

i) issue a DMA operation to the network card, the DMA operation includes instructions to copy the virtual memory block that was indicated by the RDMA read request, and ii) to transmit the copied virtual memory block to the client that submitted the RDMA read request.

2. The persistent caching system of claim 1, wherein the caching server is configured to serve RDMA read operations from the client directly from an operating system page cache of the caching server.

3. The persistent caching system of claim 2, wherein the caching server is configured to serve RDMA read operations from the client directly from the operating system page cache using an in-kernel soft-RDMA stack.

4. The persistent caching system of claim 2, wherein the caching server is configured to serve RDMA read operations from the client directly from the operating system page cache using hardware supported RDMA.

5. The persistent caching system of claim 1, wherein the client includes a client application and a storage library, the client application being linked to the storage library and provides the client access to the storage system.

6. The persistent caching system of claim 1, wherein the caching server is configured to create the virtual memory blocks using memory-mapped files.

7. The persistent caching system of claim 1, wherein a kernel of the caching server includes a paging subsystem which is configured to determine whether to store data included in a virtual memory block in a physical memory or on a disk of the memory-mapped file.

8. The persistent caching system of claim 7, wherein the caching server is configured to store data included the virtual memory block in a physical memory.

9. The persistent caching system of claim 7, wherein the caching server is configured to store data included in the virtual memory block on a disk of the associated memory-mapped file.

10. The persistent caching system of claim 7, wherein the paging subsystem is configured to keep more frequently accessed virtual memory blocks present in the physical memory and less frequently accessed virtual memory blocks on the disk of the memory-mapped file.

11. The persistent caching system of claim 7, wherein the paging subsystem is configured to:

i) subdivide a virtual memory block into a plurality of pages, and ii) decide whether to keep a first part of the plurality of pages in the physical memory and a second part of the plurality of pages on the disk of the memory-mapped file.

12. The persistent caching system of claim 5, wherein the storage library is configured to provide a write operation for writing data into the storage system and a read operation through RDMA for reading data from the storage system.

13. The persistent caching system of claim 5, wherein the caching server is configured to store data received from the storage library in a number of virtual memory blocks and to send a request for the kernel to page a number of virtual memory blocks out to the associated memory-mapped file.

14. The persistent caching system of claim 7, wherein the paging subsystem is configured to page at least one page of a virtual memory block being accessed by a write operation using the disk of the memory-mapped file.

15. A method for operating a persistent caching system, the method comprising:

storing, by one or more processors, a data in a virtual memory block of a caching server, wherein the virtual memory block refers to an associated memory-mapped file in a file system of the caching server, wherein the caching server is configured to serve RDMA read operations from the client directly from an operating system page cache of the caching server, and wherein the caching server is configured to serve RDMA read operations from the client directly from the operating system page cache of the caching server using an in-kernel soft-RDMA stack;

exporting, by one or more processors, an address of the virtual memory block from the caching server to a client;

in response to a request from the client that includes the address, accessing, by one or more processors, the virtual memory block through RDMA using the address; and

paging, by one or more processors, the virtual memory block being accessed by the client through RDMA using the memory-mapped files associated with the virtual memory block.

16. A non-transitory computer program product for operating a persistent caching system, the computer program product comprising:

one or more computer-readable storage media and program instructions stored on the one or more computer-readable storage media, the program instructions comprising;

program instructions to store a data in a virtual memory block of a caching server, wherein the virtual memory block refers to an associated memory-mapped file in a file system of the caching server;

program instructions that instruct the caching server to serve RDMA read operations directly from an operating system page cache of the caching server, wherein the RDMA read operations are from the client;

program instructions to export an address of the virtual memory block from the caching server to a client;

program instructions to respond to a request from the client that includes the address, by accessing the virtual memory block through RDMA using the address; and

program instructions to page the virtual memory block being accessed by the client through RDMA using the memory-mapped files associated with the virtual memory block.