Data storage system and method of storing data

• US 20020046322A1
• Filed: 10/01/2001
• Published: 04/18/2002
• Est. Priority Date: 10/14/2000
• Status: Active Grant

• Current Assignee: Globalfoundries US Incorporated (GlobalFoundries, Inc.)
• Original Assignee: International Business Machines Corporation
• Inventors: Butterworth, Henry Esmond, Nicholson, Robert Bruce, Turner, Douglas
• Granted Patent: US 6,704,839 B2
• Time in Patent Office: Days
• US Class Current: 711/114
• CPC Class Codes: G06F 11/2089 Redundant storage control functionality

Abstract

A data storage system and method of storing data with at least two controllers (200, 300) which share a storage space (244, 344) on an array of storage devices and which support the same set of logical upstream devices. The controllers (200, 300) share the workload by dividing the shared storage space (244, 344) into stripes where the stripes are sufficiently small to divide the workload uniformly across the storage device. This is achieved in the case of two controllers (200, 300) of a log structured array by dividing the storage space into odd and even tracks, all odd tracks being designated to one controller (300) and all even tracks to the other controller (200).

Claims

1. A data storage system comprising at least two controllers (200, 300) and a storage device with data storage space (244, 344) which is shared by the controllers (200, 300), wherein the controllers (200, 300) share the workload by dividing the shared storage space into n sets of stripes where the space in each set of stripes is designated to one controller and the stripes are sufficiently small to divide the workload uniformly across the storage device.

2. A data storage system as claimed in claim 1, wherein each controller (200, 300) manages the data in its designated stripes.

3. A data storage system as claimed in claim 1, wherein n is equal to the number of controllers (200, 300).

4. A data storage system as claimed in claim 1, wherein the units of the stripes are sufficiently small so that each portion of a host workload spans multiple stripes.

5. A data storage system as claimed in claim 1, wherein there are two controllers (200, 300) and the shared storage space (244, 344) is divided into stripes of odd and even tracks, all odd tracks being processed by one controller (300) and all even tracks being processed by the other controller (200).

6. A data storage system as claimed in claim 1, wherein the storage system includes a processor and memory, and the data storage device is an array of storage devices (106) having a plurality of data blocks organized on the storage devices in segments distributed across the storage devices, wherein when a data block in a segment stored on the storage devices in a first location is updated, the updated data block is assigned to a different segment, written to a new storage location, and designated as a current data block, and the data block in the first location is designated as an old data block, and having a main directory, stored in memory, containing the locations of the storage devices of the current data blocks.

7. A data storage system as claimed in claim 6, wherein the data storage system is a log structured array (104) and the storage device is a plurality of direct access storage devices (106).

8. A data storage system as claimed in claim 7, wherein the log structured array (104) uses check data in a storage device of an array of direct access storage devices.

9. A data storage system as claimed in claim 6, wherein write operations are mirrored to the other, or at least one other, controller (200, 300) for redundancy.

10. A data storage system as claimed in claim 6, wherein each controller (200, 300) has a primary cache (224, 324) for the data from stripes designated to that controller (200, 300) and a secondary cache (226, 326) for data from stripes designated to another controller (200, 300).

11. A data storage system as claimed in claim 6, wherein if one controller (108, 110) fails then another controller (108, 110) takes over the entire workload keeping the data structures separate so that the workload can be moved back when the failing controller (108, 110) is repaired.

12. A data storage system as claimed in claim 1, wherein each controller (200, 300) has a directory (234, 334) providing location information for data in stripes designated to that controller (200, 300).

13. A data storage system as claimed in claim 1, wherein free space collection is carried out separately by each controller (200, 300) for data in stripes designated to that controller (200, 300).

14. A data storage system as claimed in claim 1, wherein there is no contention between the controllers (200, 300) for access to the storage space (244, 344), the directories or the meta-data and no locking is required.

16. A method of storing data as claimed in claim 15, wherein each controller (200, 300) manages the data in its designated stripes.

17. A method of storing data as claimed in claim 15, wherein n is equal to the number of controllers (200, 300).

18. A method of storing data as claimed in claim 15, wherein the units of the stripes are sufficiently small so that each portion of a host workload spans multiple stripes.

19. A method of storing data as claimed in claim 15, wherein there are two controllers (200, 300) and the method comprises dividing the shared storage space (244, 344) into stripes of odd and even tracks and processing all odd tracks by one controller (300) and processing all even tracks by the other controller (200).

20. A method of storing data as claimed in claim 15, wherein the storage system includes a processor and memory, and the data storage device is an array of storage devices (106) having a plurality of data blocks organized on the storage devices in segments distributed across the storage devices, wherein when a data block in a segment stored on the storage devices in a first location is updated, the updated data block is assigned to a different segment, written to a new storage location, and designated as a current data block, and the data block in the first location is designated as an old data block, and having a main directory, stored in memory, containing the locations of the storage devices of the current data blocks.

21. A method of storing data as claimed in claim 20, wherein the shared storage space (244, 344) is a log array (104) with a plurality of direct access storage devices (106).

22. A method of storing data as claimed in claim 21, wherein the log structured array (104) uses check data in a storage device of an array of direct access storage devices.

23. A method of storing data as claimed in claim 15, wherein write operations are mirrored to the other, or at least one other, controller (200, 300) for redundancy.

24. A method of storing data as claimed in claim 15, wherein each controller (200, 300) has a primary cache (224, 324) for the data from stripes designated to that controller (200, 300) and a secondary cache (226, 326) for data from stripes designated to another controller (200, 300).

25. A method of storing data as claimed in claim 15, wherein if one controller (108, 110) fails then another controller (108, 110) takes over the entire workload keeping the data structures separate so that the workload can be moved back when the failing controller (108, 110) is repaired.

26. A method of storing data as claimed in claim 15, wherein location information is provided in each controller (200, 300) by a directory (234, 334) for data in stripes designated to that controller (200, 300).

27. A method of storing data as claimed in claim 15, wherein free space collection is carried out separately by each controller (200, 300) for data in stripes designated to that controller (200, 300).

28. A method of storing data as claimed in claim 15, wherein there is no contention between the controllers (200, 300) for access to the storage space (244, 344), the directories or meta-data and no locking is required.

15. A method of storing data in a system in which at least two controllers (200, 300) share storage space (244, 344) comprising dividing the shared storage space (244, 344) into n sets of stripes where the space of each stripe is designated to one controller (200, 300), wherein the stripes are sufficiently small to divide the workload uniformly across the storage space (244, 344).