Distributed raid storage system

• US 6,151,659 A
• Filed: 12/22/1997
• Issued: 11/21/2000
• Est. Priority Date: 12/22/1997
• Status: Expired due to Term

• Current Assignee: Emc IP Holding Company LLC (Dell Technologies Inc.)
• Original Assignee: EMC Corporation (Dell Technologies Inc.)
• Inventors: Blakeslee, Jeffrey A., Solomon, Robert C.
• Primary Examiner(s): Nguyen, Hiep T.
• Application Number:
  US08/996,444
• Time in Patent Office: 1,065 Days
• Field of Search: 711/114, 711/152, 711/148, 714/6, 710/200
• US Class Current: 711/114
• CPC Class Codes: G06F 11/1076 Parity data used in redundant arrays of independent storages, e.g. in RAID systems
• CPC Class Codes: G06F 2211/1028 Distributed, i.e. distributed RAID systems with parity
• CPC Class Codes: G06F 2211/1054 Parity-fast hardware, i.e. dedicated fast hardware for RAID systems with parity
• CPC Class Codes: G06F 3/0622 in relation to access
• CPC Class Codes: G06F 3/0637 Permissions
• CPC Class Codes: G06F 3/0689 Disk arrays, e.g. RAID, JBOD

Abstract

A data storage system has at least three data storage disks and a plurality of processing nodes in communication with the data storage disks. Each of the processing nodes shares access to the data storage disks, and each of the processing nodes includes a distributed lock manager that allows or denies access to selected stripes of data storage sectors on any of the data storage disks. Each of the processing nodes includes an interface to a private communication link to a single one of a plurality of host operating systems.

Claims

1. A data storage system comprising:

at least three data storage disks, each of said disks having a plurality of data storage sectors wherein said at least three data storage disks store data according to a protocol that forms stripes, each stripe including a plurality of data storage sectors, each of the data storage sectors in the stripe being on a different one of said data storage disks wherein one of the data storage sectors in the stripe stores parity data so that in the event that one of said different data storage disks fails, the data of said failed disk can be reconstructed from the data in the sectors of the stripe in the other of the data storage disks;

a plurality of processing nodes in communication with said at least three data storage disks, each processing node sharing access to said at least three data storage disks, and each of said processing nodes including a distributed lock manager to allow or deny access to selected stripes of data storage sectors on any of said at least three data storage disks; and

each of said processing nodes including an interface to a private communication link to a single one of a plurality of host operating systems.

2. The data storage system of claim 1 wherein said plurality of processing nodes are in communication with each other.

3. The data storage system of claim 1 wherein each distributed lock manager includes means for acquiring a lock associated with a stripe and the means for acquiring a lock includes means for awaiting release of a lock on said stripe imposed by an other of said processing nodes.

4. The data storage system of claim 1 wherein at least one of said processing nodes comprises a host-bus adapter.

5. The data storage system of claim 1 further comprising a host computer housing one of said host operating systems and at least one of said processing nodes.

6. The data storage system of claim 1 further comprising a dual loop Fibre Channel connecting said plurality of processing nodes and said at least three data storage disks.

7. The data storage system of claim 1 wherein said at least three data storage disks comprise programmable XOR drives.

8. The data storage system of claim 1 wherein each distributed lock manager includes means for asserting a NULL request for maintaining a reference to the stripe of data storage sectors without blocking other lock requests to such sectors.

9. The data storage system of claim 1 wherein each distributed lock manager includes means for asserting an unprotected read lock type of lock request for allowing nonexclusive read or write access to the stripe of data storage sectors so as to block exclusive lock requests.

10. The data storage system of claim 1 wherein each distributed lock manager includes means for asserting an unprotected write lock type of lock request for allowing nonexclusive read or write access to the stripe of data storage sectors while blocking protected and exclusive lock requests for such sectors.

11. The data storage system of claim 1 wherein each distributed lock manager includes means for asserting a protected read lock type of lock request for allowing nonexclusive read access to the stripe of data storage sectors while blocking write access and exclusive lock requests to such sectors.

12. The data storage system of claim 1 wherein each distributed lock manager includes means for asserting a protected write lock type of lock request for allowing concurrent unprotected read accesses to the stripe of data storage sectors while blocking all other lock requests.

13. The data storage system of claim 1 wherein each distributed lock manager includes means for asserting an exclusive lock type of lock request for obtaining exclusive access to the stripe of data storage sectors while blocking all other lock requests for such sectors.

14. The data storage system of claim 1 wherein each processing node further includes means for determining whether an operation on a designated sector of a designated disk has been prevented by a lock issued by any of said processing nodes with respect to the designated sector in the designated disk;

and means for awaiting release of such a lock before performing the operation on the designated sector.

15. The data storage system of claim 1 wherein said pluarlity of processing nodes comprises at least three processing nodes.

16. A method of providing each of a plurality of host operating systems access to each of a plurality of data storage disks having data stored therein according to a protocol so that in the event that one of said disks fails, the data in said failed disk can be reconstructed from the data in other ones of said disks, said method comprising:

providing a plurality of processing nodes each having a distributed lock manager and each in communication with each of the plurality of data storage disks;

providing a private communication link between each of said processing nodes and one of said host operating systems;

communicating over one of the private communication links between one of said host operating systems and one of the processing nodes;

acquiring a shared lock encompassing a stripe extending across at least three of said data storage disks;

writing new data into a selected sector in the stripe on one of said at least three data storage disks;

updating a parity sector in the stripe on another of said at least three data storage disks; and

releasing the shared lock.

17. The method of claim 16 wherein the shared lock extends over the entirety of each of said at least three data storage disks.

18. A data storage system comprising:

at least three data storage disks, each of said disks having a plurality of data storage sectors wherein said at least three data storage disks store data according to a protocol that forms stripes, each stripe including a plurality of data storage sectors, each of the data storage sectors in the stripe being on a different one of said data storage disks wherein one of the data storage sectors in the stripe stores parity data so that in the event that one of said different data storage disks fails, the data of said failed disk can be reconstructed from the data in the sectors of the stripe in the other of the data storage disks;

a plurality of processing nodes in communication with said at least three data storage disks, each of said processing nodes having a multi-leveled distributed lock manager to allow or deny access to selected stripes on the data storage sectors of any of said data storage disks; and

each of said processing nodes including an interface to a private communication link to a single one of a plurality of host operating systems.

19. The data storage system of claim 18, wherein the multi-level lock manager includes means for asserting a NULL request for maintaining a reference to the stripe of data storage sectors without blocking other lock requests to such sectors.

20. The data storage system of claim 18, wherein the multi-level lock manger includes means for asserting an unprotected read lock type of lock request for allowing nonexclusive read or write access to the stripe of data storage sectors so as to block exclusive lock requests.

21. The data storage system of claim 18, wherein the multi-level lock manager includes means for asserting an unprotected write lock type of lock request for allowing nonexclusive read or write access to the stripe of data storage sectors while blocking protected and exclusive lock requests for such sectors.

22. The data storage system of claim 18, wherein the multi-level lock manager includes means for asserting a protected read lock type of lock request for allowing nonexclusive read access to the stripe of data storage sectors while blocking write access and exclusive lock requests to such sectors.

23. The data storage system of claim 18, wherein the multi-level lock manager includes means for asserting a protected write lock type of lock request for allowing concurrent unprotected read accesses to the stripe of data storage sectors while blocking all other lock requests.

24. The data storage system of claim 18, wherein the multi-level lock manager includes means for asserting an exclusive lock type of lock request for obtaining exclusive access to the stripe of data storage sectors while blocking all other lock requests for such sectors.