Conditional storage object deletion

• US 9,002,805 B1
• Filed: 12/14/2012
• Issued: 04/07/2015
• Est. Priority Date: 12/14/2012
• Status: Expired due to Fees

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Gattu, Praveen Kumar, Denny-Brown, Derek Ernest II, Barber, Jeffrey Michael, Elving, Christopher Henning, Perry, Carl Yates
• Primary Examiner(s): Jami, Hares
• Application Number:
  US13/715,867
• Time in Patent Office: 844 Days
• Field of Search: 707609-639, 707/692
• US Class Current: 707/692
• CPC Class Codes: G06F 16/215 Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors

Abstract

Methods and apparatus for conditional deletes of storage objects are disclosed. A storage medium comprises program instructions that when executed, implement a metadata node of a storage service in which a protocol based on sequence numbers is used to resolve update conflicts. The instructions store, as part of a conditional deletion record associated with a key of a particular storage object identified as a deletion candidate, a deletion sequence number derived from a particular modification sequence number of the object. In accordance with the protocol, the instructions determine whether an additional modification sequence number larger than the deletion sequence number has been generated in response to an operation associated with the key. If such an additional sequence number has been generated, the deletion of the storage object is canceled.

Claims

1. A system, comprising one or more computing devices configured to:

implement a distributed multi-tenant storage service, in which a protocol based at least in part on respective sequence numbers associated with modification operations is used to resolve update conflicts associated with storage objects that are replicated;

store, as part of a conditional deletion record associated with a key identifying a particular storage object of the storage service, wherein the particular storage object has been identified as a candidate for removal from the storage service, a deletion sequence number derived at least in part from a particular modification sequence number associated with the particular storage object;

determine, in accordance with the protocol, whether an additional modification sequence number larger than the deletion sequence number has been generated in response to a client request to store a value associated with the key;

in response to a determination that an additional modification sequence number has been generated in response to a client request, cancel the removal of the particular storage object from the storage service.

2. The system as recited in claim 1, wherein the particular modification sequence number is based at least in part on a creation timestamp of the particular storage object.

3. The system as recited in claim 1, wherein the service comprises a sequence number generator configured to generate, in response to successive modification requests from clients, sequence numbers that differ from each other by at least a first delta quantity, and wherein the deletion sequence number is determined based at least in part on an addition of a second delta quantity to the particular modification sequence number, wherein the second delta quantity is smaller than the first delta quantity.

4. The system as recited in claim 1, wherein the one or more computing devices are further configured to:

identify the particular storage object as a candidate for removal from the storage service in accordance with on an object retention policy based at least in part on one or more of;

a creation time of the particular storage object, a most recent modification time of the particular storage object, a most recent access time of the particular storage object, or a frequency of access of the particular storage object.

5. The system as recited in claim 1, wherein the removal of the particular storage object comprises at least a first operation to delete a metadata entry identified by the key, and a second operation, asynchronous with respect to the first operation, to release storage space used for a value associated with the key.

6. A method, comprising:

implementing a distributed multi-tenant storage service, in which a protocol based at least in part on respective sequence numbers associated with modification operations is used to resolve update conflicts associated with storage objects of the storage service;

storing, as part of a conditional deletion record associated with a key identifying a particular storage object of the storage service, wherein the particular storage object has been identified as a candidate for removal from the storage service, a deletion sequence number derived at least in part from a particular modification sequence number associated with the particular storage object;

determining, in accordance with the protocol, whether an additional modification sequence number larger than the deletion sequence number has been generated in response to an operation associated with the key;

in response to determining that an additional modification sequence number has been generated, canceling the removal of the particular storage object from the storage service.

7. The method as recited in claim 6, wherein the particular modification sequence number is based at least in part on a creation timestamp of the particular storage object.

8. The method as recited in claim 6, wherein the service comprises a sequence number generator configured to generate, in response to successive modification requests from clients, sequence numbers that differ from each other by at least a first delta quantity, and wherein the deletion sequence number is determined based at least in part on an addition of a second delta quantity to the particular modification sequence number, wherein the second delta quantity is smaller than the first delta quantity.

9. The method as recited in claim 6, further comprising:

identifying the particular storage object as a candidate for removal from the storage service in accordance with on an object retention policy based at least in part on one or more of;

a creation time of the particular storage object, a most recent modification time of the particular storage object, a most recent access time of the particular storage object, or a frequency of access of the particular storage object.

10. The method as recited in claim 6, wherein the removal of the particular storage object comprises at least a first operation to delete a metadata entry identified by the key, and a second operation, asynchronous with respect to the first operation, to release storage space used for a value associated with the key.

11. The method as recited in claim 6, wherein the storage service comprises a plurality of replicas of the particular storage object maintained in accordance with a durability specification, wherein each replica has a respective metadata node, further comprising:

initiating, in accordance with the protocol, a reconciliation operation in response to a communication from a first metadata node to a second metadata node.

12. The method as recited in claim 11, further comprising:

selecting, by the first metadata node, the second metadata node as a target of the communication based at least in part on a random node selection.

13. The method of claim 6, further comprising:

identifying, by a deletion job object generator, the particular storage object as a candidate for removal from the storage service;

wherein said storing the deletion sequence number is performed asynchronously with respect to said identifying.

14. The method of claim 6, further comprising:

receiving, from a client via a programmatic interface, an indication of an object expiration policy to be used to identify candidate storage objects for removal from the storage service;

including a representation of the object expiration policy within an additional storage object (a) created in response to receiving the indication and (b) stored within the storage service.

15. A non-transitory computer-accessible storage medium storing program instructions that when executed on one or more processors:

implement a metadata node of a plurality of metadata nodes of a distributed multi-tenant storage service in which a protocol based at least in part on respective sequence numbers associated with modification operations is used to resolve update conflicts associated with storage objects of the storage service;

store, as part of a conditional deletion record associated with a key identifying a particular storage object of the storage service, wherein the particular storage object has been identified as a candidate for removal from the storage service, a deletion sequence number derived at least in part from a particular modification sequence number associated with the particular storage object;

determine, in accordance with the protocol, whether an additional modification sequence number larger than the deletion sequence number has been generated in response to an operation associated with the key; and

in response to a determination that an additional modification sequence number has been generated, cancel the removal of the particular storage object from the storage service.

16. The non-transitory computer-accessible storage medium as recited in claim 15, wherein the particular modification sequence number is based at least in part on a creation timestamp of the particular storage object.

17. The non-transitory computer-accessible storage medium as recited in claim 15, wherein the instructions when executed on the one or more processors, in response to successive modification requests from clients, generate sequence numbers that differ from each other by at least a first delta quantity, and wherein the deletion sequence number is determined based at least in part on an addition of a second delta quantity to the particular modification sequence number, wherein the second delta quantity is smaller than the first delta quantity.

18. The non-transitory computer-accessible storage medium as recited in claim 17, wherein the deletion sequence number is generated in response to a deletion operation requested by a particular worker thread of the storage service, wherein the instructions when executed on the one or more processors:

generate, in response to a second deletion operation requested by a different worker thread of the storage service, wherein the second deletion operation is associated with the key identifying the particular storage object, a second deletion sequence number equal to the deletion sequence number generated in response to the deletion operation requested by the particular worker thread.

19. The non-transitory computer-accessible storage medium as recited in claim 18, wherein the second deletion sequence number is stored in a second conditional deletion record, wherein the instructions when executed on the one or more processors:

compare, in accordance with the protocol, the respective deletion sequence numbers generated for the particular worker thread and the different worker thread; and

in response to a determination that the deletion sequence numbers are equal, discard the second conditional deletion record.

20. The non-transitory computer-accessible storage medium as recited in claim 15, wherein the particular storage object is identified as a candidate for removal from the storage service in accordance with a deletion criterion based at least in part on one or more of:

a creation time of the particular storage object, a most recent modification time of the particular storage object, a most recent access time of the particular storage object, or a frequency of access of the particular storage object.

21. The non-transitory computer-accessible storage medium as recited in claim 15, wherein the removal of the particular storage object comprises at least a first operation to delete a metadata entry identified by the key, and a second operation, asynchronous with respect to the first operation, to release storage space used for a value associated with the key.

22. The non-transitory computer-accessible storage medium as recited in claim 15, wherein the metadata node is configured to store a first replica of a plurality of replicas of a metadata entry for the particular storage object, wherein the instructions when executed on the one or more processors:

communicate with a second metadata node selected from one or more other metadata nodes storing replicas of the metadata entry to initiate a reconciliation operation in accordance with the protocol.

23. The non-transitory computer-accessible storage medium as recited in claim 22, wherein the instructions when executed on the one or more processors:

initiate the reconciliation operation in response to at least one of;

(a) a determination that a particular number of metadata entry accesses have occurred since a previous reconciliation operation, or (b) a determination that a specified amount of time has passed since a previous reconciliation operation.

24. The non-transitory computer-accessible storage medium as recited in claim 22, wherein the instructions when executed on the one or more processors:

initiate a respective reconciliation operation with each of the one or more other metadata nodes.

25. The non-transitory computer-accessible storage medium as recited in claim 15, wherein the instructions when executed on the one or more processors:

store a deletion policy metadata entry comprising a key identifying an additional storage object created in response to receiving an indication of a deletion policy from a client via a programmatic interface.