Burst buffer appliance with operating system bypass functionality to facilitate remote direct memory access

Patent metadata

Field	Value
Publication number	US-9336166-B1
Application number	US-201313905310-A
Country	US
Kind code	B1
Filing date	May 30, 2013
Priority date	May 30, 2013
Publication date	May 10, 2016
Grant date	May 10, 2016

Abstract

A burst buffer appliance is adapted for coupling between a computer system and a file system. The burst buffer appliance comprises a first memory, at least one additional memory, and a controller. The controller is configured to detect a remote direct memory access operation between a given compute node of the computer system and the additional memory, and to execute the detected remote direct memory access operation in a manner that bypasses utilization of an operating system associated with the first memory. The first and additional memories may comprise respective main and flash memories. The controller may comprise a virtual machine arranged between a host channel adaptor of the burst buffer appliance and the additional memory, with the virtual machine configured to identify a remote direct memory access operation directed to an address space of the additional memory and to control the flow of data associated with that operation.

First claim

What is claimed is:

1. An apparatus comprising:
a burst buffer appliance adapted for coupling between a computer system and a file system;
the burst buffer appliance comprising:
a first memory;
at least one additional memory; and
a controller;
the first memory and the at least one additional memory comprising respective portions of a high-speed memory of the burst buffer appliance with the high-speed memory having a substantially lower access time than the file system;
the controller comprising a remote direct memory access module of the burst buffer appliance;
the controller being configured to detect a remote direct memory access operation between a given compute node of the computer system and the additional memory, and to execute the detected remote direct memory access operation in a manner that bypasses utilization of an operating system associated with the first memory;
wherein the operating system associated with the first memory comprises a primary operating system of the burst buffer appliance;
wherein bypassing utilization of the operating system associated with the first memory comprises moving data from a host channel adaptor of the remote direct memory access module to the additional memory without utilizing computational resources of a central processing unit on which the primary operating system executes;
wherein the controller comprises a virtual machine arranged between the host channel adaptor and the additional memory;
wherein the virtual machine is configured to identify a remote direct memory access operation directed to an address space of the additional memory and to control flow of data associated with that operation between the host channel adaptor and the additional memory; and
wherein the virtual machine comprises a buffer configured to store data passing between the host channel adaptor and the additional memory, and a flow control mechanism associated with the buffer.

2. The apparatus of claim 1 wherein the first memory comprises a main memory of the burst buffer appliance.

3. The apparatus of claim 1 wherein the additional memory comprises a flash device of the burst buffer appliance.

4. The apparatus of claim 3 wherein the flash device comprises at least one flash PCIe card installed in the burst buffer appliance.

5. The apparatus of claim 3 wherein the flash device is configured to operate as a direct memory access target.

6. The apparatus of claim 3 wherein the controller comprises hardware acceleration circuitry implemented in the flash device.

7. The apparatus of claim 6 wherein the hardware acceleration circuitry implemented in the flash device comprises a buffer and associated direct memory access target circuitry.

8. The apparatus of claim 1 wherein the host channel adaptor comprises at least one of an InfiniBand host channel adaptor and a gigabit Ethernet host channel adaptor.

9. A processing platform comprising the apparatus of claim 1 .

10. The apparatus of claim 1 wherein the high-speed memory of the burst buffer appliance stores data from the computer system that is to be delivered to the file system and data from the file system that is to be delivered to the computer system.

11. The apparatus of claim 1 wherein the high-speed memory of the burst buffer appliance serves as a buffer between the computer system and the file system for storing bursts of data associated with different types of input-output operations involving the computer system and the file system.

12. A method comprising:
configuring a burst buffer appliance adapted for coupling between a computer system and a file system to include a first memory and at least one additional memory;
detecting a remote direct memory access operation between a given compute node of a computer system and the additional memory; and
executing the detected remote direct memory access operation in a manner that bypasses utilization of an operating system associated with the first memory;
the first memory and the at least one additional memory comprising respective portions of a high-speed memory of the burst buffer appliance with the high-speed memory having a substantially lower access time than the file system;
wherein the operating system associated with the first memory comprises a primary operating system of the burst buffer appliance; and
wherein bypassing utilization of said operating system associated with the first memory comprises moving data from a host channel adaptor of the burst buffer appliance to the additional memory without utilizing computational resources of a central processing unit on which the primary operating system executes;
the method further comprising:
providing a virtual machine arranged between the host channel adaptor and the additional memory;
wherein the detecting and executing are implemented at least in part utilizing the virtual machine;
wherein the detecting comprises identifying in the virtual machine a remote direct memory access operation directed to an address space of the additional memory, and further wherein the executing comprises utilizing the virtual machine to control flow of data associated with that operation between the host channel adaptor and the additional memory; and
wherein the virtual machine comprises a buffer configured to store data passing between the host channel adaptor and the additional memory, and a flow control mechanism associated with the buffer.

13. A computer program product comprising a processor-readable storage medium having encoded therein executable code of one or more software programs, wherein the one or more software programs when executed in at least one processing device cause the processing device to perform the method of claim 12 .

14. The method of claim 12 wherein the first memory comprises a main memory of the burst buffer appliance.

15. The method of claim 12 wherein the additional memory comprises a flash device of the burst buffer appliance.

16. The method of claim 15 wherein the flash device comprises at least one flash PCIe card installed in the burst buffer appliance.

17. The method of claim 15 wherein the flash device is configured to operate as a direct memory access target.

18. An information processing system comprising:
a computer system;
a file system; and
a burst buffer appliance coupled between the computer system and the file system, the burst buffer appliance comprising:
a first memory;
at least one additional memory; and
a controller;
the first memory and the at least one additional memory comprising respective portions of a high-speed memory of the burst buffer appliance with the high-speed memory having a substantially lower access time than the file system;
the controller comprising a remote direct memory access module of the burst buffer appliance;
the controller being configured to detect a remote direct memory access operation between a given compute node of the computer system and the additional memory, and to execute the detected remote direct memory access operation in a manner that bypasses utilization of an operating system associated with the first memory;
wherein the operating system associated with the first memory comprises a primary operating system of the burst buffer appliance;
wherein bypassing utilization of the operating system associated with the first memory comprises moving data from a host channel adaptor of the remote direct memory access module to the additional memory without utilizing computational resources of a central processing unit on which the primary operating system executes;
w

Assignees

• Emc Corp

Inventors

• Faibish Sorin
• Ting Dennis
• Forecast John
• Bent John M

Classifications

• G06F13/28 Primary

  using burst mode transfer, e.g. direct memory access {DMA}, cycle steal (G06F13/32 takes precedence) · CPC title

• G06F12/0246

  in block erasable memory, e.g. flash memory · CPC title