Roll back of scaled-out data

• US 10,210,190 B1
• Filed: 09/16/2013
• Issued: 02/19/2019
• Est. Priority Date: 09/16/2013
• Status: Expired due to Fees

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Long, Brian Gregory, Pfifer, Justin Thomas, Chong, Sunjae
• Primary Examiner(s): Bromell, Alexandria Y
• Application Number:
  US14/028,446
• Time in Patent Office: 1,982 Days
• Field of Search: 707695, 707696, 707741, 707684, 707711, 707737, 709201, 714 48, 714100
• CPC Class Codes: G06F 16/00 Information retrieval; Database structures therefor; File system structures therefor
• CPC Class Codes: G06F 16/22 Indexing; Data structures therefor; Storage structures
• CPC Class Codes: G06F 16/2228 Indexing structures

Abstract

A computing resource monitoring service detects an issue with processing of metrics data. The metrics data is stored in observation journals within an observation journal logical data container. A computing resource service provider may configure a map reduce processor to select a prior version of an index file, the index file comprising a mapping of the metrics data stored within a metric store and a manifest comprising a listing of all processed observation journals. The map reduce processor may identify newly unprocessed observation journals and process these journals in order to generate new data objects and a new version of an index. Subsequently, an index pointer may be modified such that obsolete versions of the index are removed and the map reduce processor refers to a different version of the index file in order to support future roll back of metrics data.

Claims

1. A computer-implemented method, comprising:

using a map reduce process for;

aggregating observation data associated with a plurality of resources that are generating resource metrics of a plurality of observation types over a period of time, the observation data being aggregated based at least in part on a corresponding subset of the resource metrics having a common type of the plurality of observation types;

generating a plurality of data segments from the observation data, each data segment of the plurality of data segments corresponding to a different time interval of the period time to which a subset of the observation data corresponds;

identifying an original index corresponding to the type, the original index comprising a manifest indicating a first set of data objects associated with the type; and

processing the manifest and the plurality of data segments to generate a second set of data objects comprising a subset of the first set of data objects, the plurality of data segments, and a new index including references to the second set of data objects;

storing the second set of data objects;

updating the original index by appending at least a portion of the new index; and

replaying the map reduce process using the original index to mitigate corrupted or defective metrics data associated with the observation data.

2. The computer-implemented method of claim 1, wherein processing the manifest and the plurality of data segments further includes arranging corresponding observation data associated with the plurality of data segments based at least in part on the resource metrics.

3. The computer-implemented method of claim 1, further comprising processing a query for one or more observations associated with the resource metrics based at least in part on a portion of the updated index prior to appending the portion of the new index.

4. The computer-implemented method of claim 3, wherein processing the query is performed using only the appended portion of the new index.

5. The computer-implemented method of claim 1, further comprising detecting one or more issues with the observation data, the observation data having been processed to generate a third set of data objects prior to aggregating the observation data.

6. The computer-implemented method of claim 1, wherein the resource metrics include observations obtained from one or more computing resources provided by a computing resource service provider to one or more customers of the computing resource service provider and includes at least processor allocation, memory allocation or network performance.

7. The computer-implemented method of claim 1, wherein the index, prior to appending the new index, includes one or more index partitions, each index partition of the one or more index partitions including a database file comprising data storage locations for at least a portion of the resource metrics.

8. A computing resource monitoring system, comprising:

memory that includes executable instructions; and

one or more processors, the one or more processors that, when executing the executable instructions, at least;

use a map reduce process that;

aggregates observation data associated with a plurality of resources that are generating resource metrics of a plurality of observation types over a period of time, the observation data being aggregated based at least in part on a corresponding subset of the resource metrics having a common type of the plurality of observation types;

generates a plurality of data segments from the observation data, each data segment of the plurality of data segments corresponding to a different time interval of the period time to which a subset of the observation data corresponds;

identifies an original index corresponding to the type, the original index comprising a manifest indicating a first set of data objects associated with the type; and

processes the manifest and the plurality of data segments to generate a second set of data objects comprising a subset of the first set of data objects, the plurality of data segments, and a new index including references to the second set of data objects;

stores the second set of data objects;

updates the original index by appending at least a portion of the new index; and

replays the map reduce process using the original index to mitigate corrupted or defective metrics data associated with the observation data.

9. The computing resource monitoring system of claim 8, wherein the instructions that, when executed by the one or more processors, process the manifest and the plurality data segments further cause the one or more processors to arrange corresponding observation data associated with the plurality of data segments based at least in part on the resource metrics.

10. The computing resource monitoring system of claim 8, wherein the instructions, when executed by the one or more processors, further cause the one or more processors to process a query for one or more observations associated with the resource metrics based at least in part on a portion of the updated index prior to appending the portion of the new index.

11. The computing resource monitoring system of claim 10, wherein the instructions that, when executed by the one or more processors, process the query further cause the one or more processors to perform the query using only the appended portion of the new index.

12. The computing resource monitoring system of claim 8, wherein the instructions, when executed by the one or more processors, further cause the one or more processors to detect one or more issues with the observation data, the observation data having been processed to generate a third set of data objects prior to aggregating the observation data.

13. The computing resource monitoring system of claim 8, wherein the resource metrics include observations obtained from one or more computing resources provided by a computing resource service provider to one or more customers of the computing resource service provider and includes at least processor allocation, memory allocation or network performance.

14. The computing resource monitoring system of claim 8, wherein the index, prior to appending the new index, includes one or more index partitions, each index partition of the one or more index partitions including a database file comprising data storage locations for at least a portion of the resource metrics.

15. One or more non-transitory computer-readable storage media having collectively stored therein instructions that, if executed by one or more processors of a computing resource monitoring system, cause the computing resource monitoring system to:

use a map reduce process to;

aggregate observation data associated with a plurality of resources that are generating resource metrics of a plurality of observation types over a period of time, the observation data being aggregated based at least in part on a corresponding subset of the resource metrics having a common type of the plurality of observation types;

generate a plurality of data segments from the observation data, each data segment of the plurality of data segments corresponding to a different time interval of the period time to which a subset of the observation data corresponds;

identify an original index corresponding to the type, the original index comprising a manifest indicating a first set of data objects associated with the type; and

process the manifest and the plurality of data segments to generate a second set of data objects comprising a subset of the first set of data objects, the plurality of data segments, and a new index including references to the second set of data objects;

store the second set of data objects;

update the original index by appending at least a portion of the new index; and

replay the map reduce process using the original index to mitigate corrupted or defective metrics data associated with the observation data.

16. The one or more non-transitory computer-readable storage media of claim 15, wherein the instructions that, when executed by the one or more processors, process the manifest and the plurality data segments further causes the one or more processors to arrange corresponding observation data associated with the plurality of data segments based at least in part on the resource metrics.

17. The one or more non-transitory computer-readable storage media of claim 15, wherein the instructions, when executed by the one or more processors, further cause the computing resource monitoring system to process a query for one or more observations associated with the resource metrics based at least in part on a portion of the updated index prior to appending the portion of the new index.

18. The one or more non-transitory computer-readable storage media of claim 17, wherein the instructions, when executed by the one or more processors, further cause the computing resource monitoring system to process the query further perform the query using only the appended portion of the new index.

19. The one or more non-transitory computer-readable storage media of claim 15, wherein the instructions that, when executed by the one or more processors, further cause the computing resource monitoring system to detect one or more issues with the observation data, the observation data having been processed to generate a third set of data objects prior to aggregating the observation data.

20. The one or more non-transitory computer-readable storage media of claim 15, wherein the resource metrics include observations obtained from one or more computing resources provided by a computing resource service provider to one or more customers of the computing resource service provider and includes at least processor allocation, memory allocation or network performance.

21. The one or more non-transitory computer-readable storage media of claim 15, wherein the index, prior to appending the new index, includes one or more index partitions, each index partition of the one or more index partitions including a database file comprising data storage locations for at least a portion of the resource metrics.