Granted Patent B2

US 11,669,545 · App. 17/107,393 · Granted Jun 6, 2023

Inventors: Jehuda Shemer (Kfar Saba, IL); Kfir Wolfson (Beer Sheva, IL); Itay Azaria (Beer Sheva, IL)

Assignee: EMC IP HOLDING COMPANY LLC

G06F16/27G06F16/215

Abstract

Claims (33)

1. A method for replicating data, the method comprising:

replicating transactions made to a production system to a storage bucket in a cloud, the transactions including writes to the production system;

generating a metadata stream associated with the transactions, wherein each entry in the metadata stream corresponds to a transaction stored in the storage bucket; and

storing the metadata stream in a metadata storage in the cloud when the metadata stream reaches a threshold size, wherein the production data is continuously replicated without maintaining a replica volume in a remote storage.

2. The method of claim 1 , wherein the transactions relate to one of files, objects, or blocks.

3. The method of claim 1 , wherein each entry in the metadata stream includes an offset and an identifier.

4. The method of claim 3 , wherein the offset refers to an offset in the production volume.

5. The method of claim 1 , further comprising consolidating the metadata stream in the metadata storage based on a consolidation interval.

6. The method of claim 1 , wherein the metadata stream includes a sequence of input/output metadata and each entry represents one of the transactions and corresponds to an input/output.

7. The method of claim 1 , further comprising capturing the transactions made to the production system and replicating the captured transaction.

8. The method of claim 7 , further comprising chunking data in the transactions into chunks, wherein the chunks are replicated to the storage bucket in the cloud in a stream that is separate from the metadata stream.

9. A non-transitory computer readable medium comprising instructions that, when executed, perform the method of claim 1 .

10. A method for rehydrating a production volume, the method comprising:

selecting a point in time in a metadata stream that corresponds to a restore point;

preparing a metadata volume by accessing the metadata stream and writing entries in the metadata stream to the metadata volume, wherein each entry includes an offset of an identifier of an object, wherein the identifier is written to a location in the metadata volume corresponding to the offset in the entry;

generating a production volume from the organized metadata volume and objects stored in a remote storage, wherein the objects in the remote storage corresponding to the identifiers stored in the organized metadata volume are written to corresponding offsets in the production volume.

11. The method of claim 10 , further comprising preparing the metadata volume prior to moving data to the production volume.

12. The method of claim 10 , wherein the production volume is only generated when a resulting replica volume based on the metadata stream and the objects stored in the remote storage is accessed like a regular volume, wherein the metadata volume prevents unnecessary reads with regard to data that will be overwritten.

13. The method of claim 10 , further comprising starting the metadata volume from a tail end of the metadata stream and populating the metadata volume with identifiers at offsets that correspond to offsets in the production volume.

14. The method of claim 13 , wherein, when generating the production volume includes using the identifier stored at the offset to retrieve an object from the remote storage.

15. The method of claim 10 , further comprising using an initialized production volume as the production volume, wherein the initialized production volume corresponds to a point in time just prior to an oldest entry in the metadata stream, wherein the oldest entry is the tail end of the metadata stream.

16. The method of claim 10 , further comprising reading the metadata stream up to the selected point in time.

17. A non-transitory storage medium having stored therein instructions that are executable by one or more hardware processors to perform operations comprising:

selecting a point in time in a metadata stream that corresponds to a restore point;

preparing a metadata volume by accessing the metadata stream and writing entries in the metadata stream to the metadata volume, wherein each entry includes an offset of an identifier of an object, wherein the identifier is written to a location in the metadata volume corresponding to the offset in the entry;

generating a production volume from the organized metadata volume and objects stored in a remote storage, wherein the objects in the remote storage corresponding to the identifiers stored in the organized metadata volume are written to corresponding offsets in the production volume.

18. The non-transitory storage medium of claim 17 , further comprising preparing the metadata volume prior to moving data to the production volume.

19. The non-transitory storage medium of claim 17 , wherein the production volume is only generated when a resulting replica volume based on the metadata stream and the objects stored in the remote storage is accessed like a regular volume, wherein the metadata volume prevents unnecessary reads with regard to data that will be overwritten.

20. The non-transitory storage medium of claim 17 , further comprising:

starting the metadata volume from a tail end of the metadata stream and populating the metadata volume with identifiers at offsets that correspond to offsets in the production volume;

using an initialized production volume as the production volume, wherein the initialized production volume corresponds to a point in time just prior to an oldest entry in the metadata stream, wherein the oldest entry is the tail end of the metadata stream;

reading the metadata stream up to the selected point in time; and

when generating the production volume, using the identifier stored at the offset to retrieve an object from the remote storage.

Assignments (9)

Continuity (2)