Granted Patent B1

US 9,626,120 · App. 14/979,628 · Granted Apr 18, 2017

Abstract

Claims (64)

1. A computer-implemented method for dynamically adjusting batch request sizes, at least a portion of the method being performed by a computing device comprising at least one processor, the method comprising:

identifying a storage system that accepts batched requests for data stored within the storage system;

measuring a latency for at least one request sent to the storage system to retrieve a portion of the data stored within the storage system;

configuring a batched request for the storage system by:

determining that a plurality of requests in the batched request are requesting data stored in a plurality of data containers within the storage system;

sorting the requests in the batched request according to the data containers the requests are requesting the data from; and

assigning the sorted requests to a plurality of threads in a multi-threaded system, wherein each request to the same data container is assigned to the same thread within the plurality of threads; and

adjusting, based on the latency of the request, a batch size for the batched request; and

sending the configured batched request to the storage system.

2. The computer-implemented method of claim 1 , wherein adjusting the batch size comprises at least one of:

increasing the batch size in response to determining that the latency of the request is higher than a previously measured latency that was previously used to set the batch size; and

decreasing the batch size in response to determining that the latency of the request is lower than a previously measured latency that was previously used to set the batch size.

3. The computer-implemented method of claim 1 , further comprising:

determining that a certain request in the batched request is expected to take longer than a predetermined time limit to return;

assigning the certain request to a subsequent batched request; and

returning all requests in the batched request except the certain request.

4. The computer-implemented method of claim 3 , wherein the certain request is expected to take longer than the predetermined time limit because the certain request is expected to pass through a plurality of layers of systems.

5. The computer-implemented method of claim 3 , wherein the certain request is expected to take longer than the predetermined time limit because the certain request shares at least one characteristic with at least one previous request that took longer than the predetermined time limit.

6. The computer-implemented method of claim 5 , further comprising recording characteristics of requests that take longer than the predetermined time in order to predict future requests that are expected take longer than the predetermined time limit.

7. The computer-implemented method of claim 1 , wherein:

the storage system comprises a backup storage system; and

the batched request comprises a batch of read requests made in order to provision a system from the backup storage system.

8. A system for dynamically adjusting batch request sizes, the system comprising:

an identification module, stored in memory, that identifies a storage system that accepts batched requests for data stored within the storage system;

a measuring module, stored in memory, that measures a latency for at least one request sent to the storage system to retrieve a portion of the data stored within the storage system;

an adjusting module, stored in memory, that configures a batched request for the storage system by:

adjusting, based on the latency of the request, a batch size for the batched request;

determining that a plurality of requests in the batched request are requesting data stored in a plurality of data containers within the storage system;

sorting the requests in the batched request according to the data containers the requests are requesting the data from; and

assigning the sorted requests to a plurality of threads in a multi-threaded system, wherein each request to the same data container is assigned to the same thread within the plurality of threads;

a sending module, stored in memory, that sends the configured batched request to the storage system; and

at least one physical processor configured to execute the identification module, the measuring module, the adjusting module, and the sending module.

9. The system of claim 8 , wherein the adjusting module adjusts the batch size by at least one of:

increasing the batch size in response to determining that the latency of the request is higher than a previously measured latency that was previously used to set the batch size; and

decreasing the batch size in response to determining that the latency of the request is lower than a previously measured latency that was previously used to set the batch size.

10. The system of claim 8 , further comprising a batch adjustment module, stored in memory, that:

determines that a certain request in the batched request is expected to take longer than a predetermined time limit to return;

assigns the certain request to a subsequent batched request; and

returns all requests in the batched request except the certain request.

11. The system of claim 10 , wherein the certain request is expected to take longer than the predetermined time limit because the certain request is expected to pass through a plurality of layers of systems.

12. The system of claim 8 , wherein the batch size comprises a time interval during which all requests received are placed into a same batch.

13. The system of claim 8 , wherein the batch size comprises a set number of requests.

14. The system of claim 8 , wherein:

the storage system comprises a backup storage system; and

the batched request comprises a batch of read requests made in order to provision a system from the backup storage system.

15. A non-transitory computer-readable medium comprising one or more computer-readable instructions that, when executed by at least one processor of a computing device, cause the computing device to:

identify a storage system that accepts batched requests for data stored within the storage system;

measure a latency for at least one request sent to the storage system to retrieve a portion of the data stored within the storage system;

configure a batched request for the storage system by:

adjusting, based on the latency of the request, a batch size for the batched requests sent to the storage system;

determining that a plurality of requests in the batched request are requesting data stored in a plurality of data containers within the storage system;

sorting the requests in the batched request according to the data containers the requests are requesting the data from; and

assigning the sorted requests to a plurality of threads in a multi-threaded system, wherein each request to the same data container is assigned to the same thread within the plurality of threads; and

send the configured batched request to the storage system.

16. The non-transitory computer-readable medium of claim 15 , wherein the one or more computer-readable instructions cause the computing device to adjust the batch size by at least one of:

increasing the batch size in response to determining that the latency of the request is higher than a previously measured latency that was previously used to set the batch size; and

decreasing the batch size in response to determining that the latency of the request is lower than a previously measured latency that was previously used to set the batch size.

17. The non-transitory computer-readable medium of claim 15 , wherein the one or more computer-readable instructions cause the computing device to:

determine that a certain request in the batched request is expected to take longer than a predetermined time limit to return;

assign the certain request to a subsequent batched request; and

return all requests in the batched request except the certain request.

18. The non-transitory computer-readable medium of claim 17 , wherein the certain request is expected to take longer than the predetermined time limit because the certain request is expected to pass through a plurality of layers of systems.

19. The non-transitory computer-readable medium of claim 15 , wherein the one or more computer-readable instructions cause the computing device to measure the latency for the at least one request by measuring the latency as a number of input/output operations per second performed on the storage system over a predetermined time period.

20. The non-transitory computer-readable medium of claim 19 , wherein the one or more computer-readable instructions cause the computing device to measure the latency for the at least one request by measuring the latency at predetermined intervals.

Assignments (11)