Storage replication system with data tracking

• US 20060182050A1
• Filed: 01/28/2005
• Published: 08/17/2006
• Est. Priority Date: 01/28/2005
• Status: Abandoned Application

• Current Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Dohm, Karl
• Application Number:
  US11/046,350
• Publication Number:
  US 20060182050A1
• Time in Patent Office: Days
• US Class Current: 370/312
• CPC Class Codes: G06F 11/2056 by mirroring
• CPC Class Codes: G06F 11/2058 using more than 2 mirrored copies
• CPC Class Codes: G06F 11/2069 Management of state, configuration or failover
• CPC Class Codes: G06F 11/2071 using a plurality of controllers
• CPC Class Codes: G06F 11/2082 Data synchronisation

Abstract

A storage replication method comprises replicating data from a source among a plurality of destinations and tracking data modifications in the destinations. Identification of the modifications is mutually communicated among multiple destination arrays. In a source failover event, a selected destination is established as a new source, reforming the replicated data in the remaining destinations into synchrony with the new source.

Claims

1. A storage replication method comprising:

replicating data from a source among a plurality of destinations;

tracking data modifications in the plurality of destinations;

mutually communicating the tracked data modifications among the plurality of destinations; and

in a source failover condition, assigning a selected destination as a new source and reforming data in remaining destinations into synchrony with the new source, the reforming being limited to data that differs from the new source.

2. The method according to claim 1 further comprising:

tracking, at individual destinations of the destination plurality, modified data blocks in a destination logical unit (lun).

3. The method according to claim 1 further comprising:

detecting, at an individual destination of the destination plurality, a write directed to a logical unit (lun) of the individual destination to which a fanout relationship exists with the source; and

sending an asynchronous communication packet to ones of the destination plurality.

4. The method according to claim 1 further comprising:

detecting, at an individual destination of the destination plurality, a write directed to a logical unit (lun) of the individual destination to which a fanout relationship exists with the source;

collecting a data packet including block numbers modified by one or more writes and sequence numbers indicating unique identifiers for block content; and

sending the data packet by asynchronous communication to ones of the destination plurality.

5. The method according to claim 4 further comprising:

combining a plurality of data packets into a packet group; and

sending the packet group by asynchronous communication to ones of the destination plurality.

6. The method according to claim 4 further comprising:

receiving, at a receiving destination of the destination plurality, a plurality of data packets and sequence numbers from ones of the destination plurality; and

determining differences in data content among ones of the destination plurality.

7. The method according to claim 1 further comprising:

detecting a failover condition;

selecting a new source from among the plurality of destinations; and

sending from the new source a signal initiating data reformation in the plurality of destinations.

8. The method according to claim 7 further comprising:

receiving, at a destination of the destination plurality, the signal initiating data reformation;

terminating processing of buffered writes pending from a previous replication write stream; and

sending, to the new source, an updated block list in the destination, the updated block list including a list of blocks updated by the replication.

9. The method according to claim 8 further comprising:

determining, at the new source, data to be sent to the destination plurality based on updated block lists from the destination plurality; and

copying data from the new source to the destination plurality, the copied data being sufficient to synchronize the new source and destination plurality.

10. The method according to claim 9 further comprising:

determining, by the new source, whether a destination of the destination plurality has a more current state than the new source;

sending from the new source to the destination having the more current state a request for data in the destination that is not current in the new source;

returning requested data from the destination having the more current state to the new source; and

updating the new source with the requested data.

11. A storage unit adapted for usage in a redundant data storage system comprising:

a storage;

an input/output interface coupled to the storage and adapted to communicate with a plurality of distributed site storage units;

a controller coupled to the storage and the input/output interface; and

a logic executable on the controller adapted to operate the storage unit as a secondary site in a fanout arrangement, replicate data to the storage from a primary site storage unit, track modifications in data written to storage, communicate the tracked modifications among the plurality of distributed site storage units, and collect tracked changes received from the plurality of distributed site storage units.

12. The storage unit according to claim 11 further comprising:

the logic adapted to receive a command to failover and, in response to the command to failover, operate the storage unit as a primary site and send a request to reform fanout to the plurality of distributed site storage units.

13. The storage unit according to claim 11 further comprising:

the logic adapted to receive a command to failover and, in response to the command to failover, operate the storage unit as a primary site, send a request to reform fanout to the plurality of distributed site storage units, determine data to be sent to the plurality of distributed site storage units based on updated block lists from the plurality of distributed site storage units, and copy data to the plurality of distributed site storage units, the copied data being sufficient to replicate data in the primary site storage unit.

14. The storage unit according to claim 11 further comprising:

the storage configured at least partly as logical units (luns); and

the logic adapted to detect a write directed to a logical unit (lun) to which a fanout relationship exists with the primary site storage unit, track modified blocks in a storage logical unit (lun), and send an asynchronous communication packet to the plurality of distributed site storage units.

15. The storage unit according to claim 14 further comprising:

the logic adapted to collect a data packet including block numbers modified by one or more writes and sequence numbers indicating unique identifiers for block content and send the data packet by asynchronous communication to the plurality of distributed site storage units.

16. The storage unit according to claim 15 further comprising:

the logic adapted to combine a plurality of data packets into a packet group and send the data packet by asynchronous communication to the plurality of distributed site storage units.

17. The storage unit according to claim 15 further comprising:

the logic adapted to receive a plurality of data packets and sequence numbers from the plurality of distributed site storage units and determine differences in data content among the plurality of distributed site storage units.

18. The storage unit according to claim 11 further comprising:

the logic adapted to receive a signal initiating data reformation, terminate processing of buffered writes pending from a previous replication write stream, and send to a storage unit newly operating as a primary site an updated block list, the updated block list including a list of blocks updated by the replication.

19. The storage unit according to claim 11 further comprising:

the logic adapted to receive a command to failover, reconfigure from operation as a secondary site storage unit to a new primary site storage unit, and send a signal informing remaining secondary site storage units of the plurality of distributed site storage units that fanout is reforming; and

the logic operable for a new primary site storage unit and adapted to;

determine whether a remaining secondary site storage unit of the plurality of distributed site storage units has a more current state than the new source;

send a request to the secondary site storage unit having the more current state for data that is not current in the new primary site storage unit; and

updating the new primary site storage unit with the requested data.

20. A storage system comprising:

a plurality of storage arrays arranged in a 1;

n fanout configuration; and

a logic executable in the plurality of storage arrays adapted to track data modifications during data replication from a source storage array to n destination storage arrays, mutually share tracked data modification information among the n destination storage arrays, and respond to a failover condition by reforming to a 1;

n−

1 fanout configuration, the reformation being directed according to the mutually shared tracked data modification information from the n destination storage arrays.

21. The storage system according to claim 20 further comprising:

the logic adapted to respond to the failover condition by configuring the plurality of storage arrays to exclude the failed source storage array and assign one of the n destination storage arrays to operate as a new source storage array in an assignment made substantially contemporaneously with the failover.

22. The storage system according to claim 21 further comprising:

the logic adapted to further respond to the failover condition by reforming data in remaining n−

1 destination storage arrays into compliance with the new source storage array.

23. The storage system according to claim 20 further comprising:

a logic executable in individual destination storage arrays adapted to track modified data blocks in a destination logical unit (lun).

24. The storage system according to claim 20 further comprising:

a logic executable in individual destination storage arrays adapted to detect a write directed to a logical unit (lun) to which a fanout relationship exists with the source storage array and adapted to send an asynchronous communication packet to others of the destination storage array plurality.

25. The storage system according to claim 20 further comprising:

a logic executable in individual destination storage arrays adapted to;

detect a write directed to a logical unit (lun) to which a fanout relationship exists with the source storage array;

collect a data packet including block numbers modified by one or more writes and sequence numbers indicating unique identifiers for block content; and

send the data packet by asynchronous communication to others of the destination storage array plurality.

26. The storage system according to claim 25 further comprising:

a logic executable in individual destination storage arrays further adapted to combine a plurality of data packets into a packet group and send the packet group by asynchronous communication to others of the destination storage array plurality.

27. The storage system according to claim 25 further comprising:

a logic executable in individual destination storage arrays further adapted to receive a plurality of data packets and sequence numbers from others of the destination storage array plurality and determine differences in data content among the destination storage array plurality.

28. The storage system according to claim 20 further comprising:

a logic executable in individual destination storage arrays further adapted to;

receive a command to failover;

reconfigure as a new source storage array; and

send a signal informing remaining destination storage arrays in the destination storage array plurality that fanout is reforming.

29. The storage system according to claim 28 further comprising:

a logic executable in individual destination storage arrays adapted to;

receive the signal informing of fanout reforming;

terminate processing of buffered writes pending from a previous replication write stream; and

send to the new source storage array an updated block list in the destination, the updated block list including a list of blocks updated by the replication.

30. The storage system according to claim 29 further comprising:

a logic executable in the new source storage array adapted to;

determine differences in updated block lists received from the destination storage array plurality; and

copy data to the destination storage array plurality sufficient to synchronize the storage array plurality.

31. The storage system according to claim 30 further comprising:

a logic executable in the new source storage array adapted to;

determine whether a destination storage array of the destination storage array plurality has a more current state than the new source storage array;

send to a destination storage array having a most current state a request for data that is present in the destination storage array and not present in the new source storage array; and

updating the new source storage array with data received in response to the request.

32. An article of manufacture comprising:

a controller usable medium having a computable readable program code embodied therein for performing storage replication, the computable readable program code further comprising;

a code capable of causing the controller to replicate data from a source among a plurality of destinations;

a code capable of causing the controller to track data modifications in the plurality of destinations;

a code capable of causing the controller to mutually communicate the tracked data modifications among the plurality of destinations; and

a code capable of causing the controller to respond to a source failover condition by assigning a selected destination as a new source and reforming data in remaining destinations into synchrony with the new source.

33. A storage replication method comprising:

replicating data from a source to a plurality of destinations;

detecting a source failover condition;

selecting a new source from among the plurality of destinations based on conditions contemporaneous with the failover condition; and

sending from the new source a signal initiating data reformation in the plurality of destinations.

34. The method according to claim 33 further comprising:

distributing replication status information for the individual destinations throughout the plurality of destinations during data replication; and

reforming data in the plurality of destinations into synchrony with the new source using the replication status information.

35. A storage replication method comprising:

replicating data from a source to a plurality of destinations;

receiving, at a destination of the destination plurality, a signal initiating data reformation;

terminating processing of buffered writes pending from a previous replication write stream; and

sending, to the new source, an updated block list in the destination, the updated block list including a list of blocks updated by the replication.

36. The method according to claim 35 further comprising:

determining, at the new source, data to be sent to the destination plurality based on updated block lists from the destination plurality; and

copying data from the new source to the destination plurality, the copied data being sufficient to synchronize the new source and destination plurality.

37. The method according to claim 35 further comprising:

determining, by the new source, whether a destination of the destination plurality has a more current state than the new source;

sending from the new source to the destination having a most current state a request for data in the destination that is not current in the new source;

returning requested data from the destination having the most current state to the new source; and

updating the new source with the requested data.

38. A storage unit adapted for usage in a redundant data storage system comprising:

means for storing data;

means coupled to the data storing means for communicating with a plurality of distributed site storage units; and

means coupled to the data storing means and to the communicating means for operating as a secondary site that replicates data from a primary site;

means for tracking modifications in replicated data;

means for communicating tracked modifications among the plurality of distributed storage units; and

means for collecting tracked changes received from the plurality of distributed storage units.

39. The storage unit according to claim 38 further comprising:

means for receiving a command to failover;

means responsive to the failover command for operating as a primary site and sending a request to reform fanout to the plurality of distributed site storage units;

means for determining data to be sent to the plurality of distributed site storage units based on updated block lists from the plurality of distributed site storage units; and

means for copying data to the plurality of distributed site storage units, the copied data being sufficient to replicate data in the primary site storage unit.

40. The storage unit according to claim 38 further comprising:

means for receiving a command to failover;

means responsive to the failover command for reconfiguring from operation as a secondary site storage unit to a new primary site storage unit;

means for informing remaining secondary site storage units of the plurality of distributed site storage units that fanout is reforming;

means for determining whether a remaining secondary site storage unit of the plurality of distributed site storage units has a more current state than the new source;

means for sending a request to the secondary site storage unit having the more current state for data that is not current in the new primary site storage unit; and

means for updating the new primary site storage unit with the requested data.