Cluster file system with a burst buffer appliance for coordinated control of data movement among storage tiers across a plurality of object storage servers

• US 9,454,331 B1
• Filed: 06/27/2013
• Issued: 09/27/2016
• Est. Priority Date: 06/27/2013
• Status: Active Grant

• Current Assignee: EMC Corporation (Dell Technologies Inc.)
• Original Assignee: EMC Corporation (Dell Technologies Inc.)
• Inventors: Bent, John M., Faibish, Sorin, Gupta, Uday, Peng, Tao
• Primary Examiner(s): Beausoliel, Jr., Robert
• Assistant Examiner(s): Pham, Michael
• Application Number:
  US13/928,940
• Time in Patent Office: 1,188 Days
• Field of Search: 707/823, 707/824
• US Class Current: 1/1
• CPC Class Codes: G06F 16/13 File access structures, e.g. distributed indices arrangements of input from, or output to, record carriers G06F3/06
• CPC Class Codes: G06F 3/061 Improving I/O performance
• CPC Class Codes: G06F 3/0643 Management of files
• CPC Class Codes: G06F 3/0659 Command handling arrangements, e.g. command buffers, queues, command scheduling
• CPC Class Codes: G06F 3/067 Distributed or networked storage systems, e.g. storage area networks [SAN], network attached storage [NAS]

Abstract

A cluster file system comprises a burst buffer appliance coupled to a plurality of object storage servers via a network. The burst buffer appliance is configured to implement storage tiering control functionality for at least first and second storage tiers comprising respective disjoint subsets of the plurality of object storage servers. The burst buffer appliance implements a coordinated movement of data between the first and second storage tiers to pre-fetch at least one additional portion of a single logical file that is stored across a plurality of said object storage devices from another of said plurality of object storage devices. The burst buffer appliance optionally implements the coordinated movement of data between the first and second storage tiers such that substantially all portions of a single logical file that are stored across a plurality of said object storage devices in said cluster file system are stored in only one of said storage tiers at a given time.

Claims

1. An object storage device of a cluster file system comprising a plurality of object storage servers and a plurality of object storage devices, comprising:

a burst buffer appliance configured to communicate with said plurality of object storage servers of the cluster file system over a network;

the burst buffer appliance further comprising;

a controller configured to (i) implement storage tiering control functionality for at least first and second storage tiers comprising respective disjoint subsets of the plurality of object storage servers to control movement of data between the first and second storage tiers; and

(ii) notify, in response to a read request for a first portion of a single logical file stored across a plurality of said object storage devices, at least one additional burst buffer appliance of at least one additional object storage device to implement a coordinated movement of data between first and second storage tiers of said at least one additional object storage device to pre-fetch at least one additional portion of said single logical file that is stored by said at least one additional object storage device;

the object storage device being implemented utilizing at least one processing device coupled to a memory.

2. The object storage device of claim 1 wherein said burst buffer appliance further notifies said at least one additional object storage device to implement said coordinated movement of data between the first and second storage tiers such that substantially all portions of said single logical file that are stored across said plurality of said object storage devices in said cluster file system are stored in only one of said storage tiers at a given time.

3. The object storage device of claim 1 further comprising a parallel log structured file system daemon for communicating with parallel log structured file system daemons on other object storage devices in said cluster file system to implement said coordinated movement of data.

4. The object storage device of claim 3 wherein said parallel log structured file system daemon notifies one or more of said parallel log structured file system daemons on other object storage devices in said cluster file system to pre-fetch portions of said single logical file that are stored across said plurality of said object storage devices.

5. The object storage device of claim 1 wherein all portions of said single logical file that are stored across said plurality of said object storage devices are stored in said first storage tier.

6. The object storage device of claim 1 wherein all portions of said single logical file that are stored across said plurality of said object storage devices are stored in said second storage tier.

7. The object storage device of claim 1 wherein each of the object storage servers is configured to interface with at least one object storage target, and further wherein the object storage servers in the first storage tier are configured to interface with object storage targets of a first type and the object storage servers in the second storage tier are configured to interface with object storage targets of a second type different than the first type.

8. The object storage device of claim 7 wherein the object storage targets of the first type comprise respective non-volatile electronic storage devices including one or more of flash memory, magneto-resistive random-access memory (MRAM) and phase-change random access memory (PC-RAM).

9. The object storage device of claim 6 wherein the object storage targets of the second type comprise respective disk storage devices.

11. The object storage device of claim 1 wherein the burst buffer appliance is further configured to communicate with a plurality of clients over the network and to process requests from said clients.

12. The object storage device of claim 1 wherein the controller comprises a data placement and migration controller.

13. The object storage device of claim 1 wherein at least a portion of the controller is implemented using data migration software at least a portion of which is stored in said memory and executed by said at least one processing device.

10. The object storage device of claim wherein the cluster file system comprises a Lustre file system.

14. A processing platform comprising an object storage device of a cluster file system comprising a plurality of object storage servers and a plurality of object storage devices, said object storage device comprising:

a burst buffer appliance configured to communicate with said plurality of object storage servers of the cluster file system over a network;

the burst buffer appliance further comprising;

a controller configured to (i) implement storage tiering control functionality for at least first and second storage tiers comprising respective disjoint subsets of the plurality of object storage servers to control movement of data between the first and second storage tiers; and

(ii) notify, in response to a read request for a first portion of a single logical file stored across a plurality of said object storage devices, at least one additional burst buffer appliance of at least one additional object storage device to implement a coordinated movement of data between first and second storage tiers of said at least one additional object storage device to pre-fetch at least one additional portion of said single logical file that is stored by said at least one additional object storage device;

the object storage device being implemented utilizing at least one processing device coupled to a memory.

15. A method comprising:

configuring a burst buffer appliance of a cluster file system comprising a plurality of object storage servers and a plurality of object storage devices to communicate with a plurality of object storage servers of the cluster file system over a network; and

implementing storage tiering control functionality in the burst buffer appliance for at least first and second storage tiers comprising respective disjoint subsets of the plurality of object storage servers to control movement of data between the first and second storage tiers; and

notifying, in response to a read request for a first portion of a single logical file stored across a plurality of said object storage devices, at least one additional burst buffer appliance of at least one additional object storage device to implement a coordinated movement of data between first and second storage tiers of said at least one additional object storage device to pre-fetch at least one additional portion of said single logical file that is stored by said at least one additional object storage device.

16. The method of claim 15 wherein said burst buffer appliance further notifies said at least one additional object storage device to implement said coordinated movement of data between the first and second storage tiers such that substantially all portions of said single logical file that are stored across said plurality of said object storage devices in said cluster file system are stored in only one of said storage tiers at a given time.

17. The method of claim 15 further comprising the step of a parallel log structured file system daemon on a first object storage device communicating with parallel log structured file system daemons on other object storage devices in said cluster file system to implement said coordinated movement of data.

18. The method of claim 17 wherein said parallel log structured file system daemon is further configured to notify one or more of said parallel log structured file system daemons on other object storage devices in said cluster file system to prefetch portions of said single logical file that are stored across a plurality of said object storage devices.

19. The method of claim 15 further comprising moving data between the first and second storage tiers under control of the burst buffer appliance.

20. The method of claim 15 further comprising:

configuring the object storage servers in the first storage tier to interface with object storage targets of a first type; and

configuring the object storage servers in the second storage tier to interface with object storage targets of a second type different than the first type;

wherein the first and second different types of object storage targets provide respective different levels of performance.

21. A computer program product comprising a non-transitory machine-readable storage medium having encoded therein executable code of one or more software programs, wherein the one or more software programs when executed cause a burst buffer appliance to perform the following steps:

configuring the burst buffer appliance of a cluster file system comprising a plurality of object storage servers and a plurality of object storage devices to communicate with a plurality of object storage servers of the cluster file system over a network;

implementing storage tiering control functionality in the burst buffer appliance for at least first and second storage tiers comprising respective disjoint subsets of the plurality of object storage servers to control movement of data between the first and second storage tiers; and

notifying, in response to a read request for a first portion of a single logical file stored across a plurality of said object storage devices, at least one additional burst buffer appliance of at least one additional object storage device to implement a coordinated movement of data between first and second storage tiers of said at least one additional object storage device to pre-fetch at least one additional portion of said single logical file that is stored by said at least one additional object storage device.

22. A cluster file system comprising:

a burst buffer appliance; and

a plurality of object storage servers coupled to the burst buffer appliance via a network;

a plurality of object storage devices;

the burst buffer appliance further comprising;

a controller configured to implement storage tiering control functionality for at least first and second storage tiers comprising respective disjoint subsets of the plurality of object storage servers;

the burst buffer appliance thereby being configured to (i) control movement of data between the first and second storage tiers; and

(ii) notify, in response to a read request for a first portion of a single logical file stored across a plurality of said object storage devices, at least one additional burst buffer appliance of at least one additional object storage device to implement a coordinated movement of data between first and second storage tiers of said at least one additional object storage device to pre-fetch at least one additional portion of said single logical file that is stored by said at least one additional object storage device.

23. The system of claim 22 wherein said burst buffer appliance further implements said coordinated movement of data between the first and second storage tiers such that substantially all portions of a single logical file that are stored across a plurality of said object storage devices in said cluster file system are stored in only one of said storage tiers at a given time.

24. The system of claim 22 further comprising a parallel log structured file system daemon for communicating with parallel log structured file system daemons on other object storage devices in said cluster file system to implement said coordinated movement of data.

25. The system of claim 24 wherein said parallel log structured file system daemon notifies one or more of said parallel log structured file system daemons on other object storage devices in said cluster file system to pre-fetch portions of said single logical file that are stored across a plurality of said object storage devices.

26. The system of claim 22 wherein the cluster file system comprises a Lustre file system.