Resource allocation for staged execution pipelining

• US 10,353,593 B2
• Filed: 03/09/2018
• Issued: 07/16/2019
• Est. Priority Date: 09/03/2013
• Status: Active Grant

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Alapati, Nishanth, Vincent, Pradeep, Salyers, David Carl
• Primary Examiner(s): Tseng, Cheng Yuan
• Application Number:
  US15/917,192
• Publication Number:
  US 20180260138A1
• Time in Patent Office: 494 Days
• Field of Search: None
• CPC Class Codes: G06F 3/06 Digital input from, or digital output to, record carriers , e.g. RAID, emulated record carriers or networked record carriers
• CPC Class Codes: G06F 3/0611 in relation to response time
• CPC Class Codes: G06F 3/0631 by allocating resources to storage systems
• CPC Class Codes: G06F 3/0653 Monitoring storage devices or systems
• CPC Class Codes: G06F 3/067 Distributed or networked storage systems, e.g. storage area networks [SAN], network attached storage [NAS]
• CPC Class Codes: G06F 9/5083 Techniques for rebalancing the load in a distributed system
• CPC Class Codes: H04L 67/01 Protocols
• CPC Class Codes: H04L 67/1097 for distributed storage of data in networks, e.g. transport arrangements for network file system [NFS], storage area networks [SAN] or network attached storage [NAS]

Abstract

A method and apparatus for staged execution pipelining and allocating resource to staged execution pipelines are provided. One or more execution pipelines are established, where each of the one or more execution pipelines includes one or more execution stages. Data is provided to the one or more execution pipelines for processing and resources are allocated to the execution pipeline.

Claims

1. A computer-implemented method, comprising:

receiving a first data processing request of a plurality of data processing requests comprising a plurality of processing types, the first data processing request comprising a first processing type of the plurality of processing types for first data associated with the first data processing request;

determining, based on the first processing type, the plurality of data processing requests, and information associated with one or more properties of a computing resource, an execution pipeline associated with the computing resource, the execution pipeline comprising a first subset of the computing resource capable of executing operations associated with the first processing type;

establishing the execution pipeline, for executing the plurality of processing types in one or more execution stages, by associating the first subset of the computing resource with the first data processing request;

causing the execution pipeline to execute the first data processing request on the first data using the first subset of the computing resource;

monitoring execution of the first data processing request in the execution pipeline; and

providing an outcome of the execution of the first data processing request.

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

receiving a second data processing request for second data having a second processing type of the plurality of processing types;

determining, based on the second processing type, a second subset of the computing resource;

further establishing the execution pipeline by associating the second subset of the computing resource with the second data processing request; and

causing the execution pipeline to execute the second data processing request on the second data using the second subset of the computing resource.

3. The computer-implemented method of claim 1, wherein the execution pipeline is provided in response to the first data processing request.

4. The computer-implemented method of claim 3, wherein the one or more properties of the computing resource include at least one of bandwidth, CPU thread or core count, local storage capacity, and local storage input/output bandwidth.

5. The computer-implemented method of claim 1, further comprising adding the first data to a first in, first out (FIFO) queue before causing the execution of the first data processing request.

6. The computer-implemented method of claim 1, further comprising causing storage, by a data storage system, of the first data after the execution of the first data processing request is complete.

7. A system, comprising:

at least one computing device implementing one or more services that;

queue a plurality of data processing requests, a data processing request of the plurality comprising a processing type for data associated with the data processing request;

determine, based on one or more properties of the plurality of data processing requests and information associated with a configuration of a computing resource, an execution pipeline associated with a subset of the computing resource capable of executing operations associated with the processing type, the one or more properties of the plurality of data processing requests comprising the processing type of the data processing request;

generate the execution pipeline to comprise one or more stages associated with a set of processing types of which the processing, type is a member by associating the subset of the computing resource with the data processing request;

execute the data processing request on the data using the execution pipeline; and

monitor execution of the data processing request in the execution pipeline to provide an outcome of the execution to a requester.

8. The system of claim 7, wherein the one or more services further retrieve, prior to execution of the data processing request, a subset of the data having a size conforming to a configuration of the execution pipeline.

9. The system of claim 7, wherein the one or more services buffer the data prior to execution of the data processing request.

10. The system of claim 7, further comprising a network interface that receives the data from the execution pipeline after execution of the data processing request and transmits the data, after execution, over a network to a data storage system.

11. The system of claim 7, wherein the computing resource is an input/output resource.

12. The system of claim 7, wherein the configuration of the computing resource includes at least one of network bandwidth, CPU thread or core count, local storage capacity, and local storage input/output bandwidth.

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

queue a plurality of data processing requests, a data processing request of the plurality of data processing requests including a processing type for data associated with the data processing request;

allocate a computing resource to an execution pipeline to process the plurality of data processing requests, a subset of the computing resource allocated to the data processing request based on the processing type, one or more properties of the plurality of data processing requests not including the data processing request, and information associated with a configuration of the computing resource, the execution pipeline comprising a subset of the computing resource capable of executing operations associated with the processing type associated with an execution stage of the execution pipeline;

execute the data processing request on the data using the execution pipeline; and

monitor execution of the data processing request in the execution pipeline.

14. The one or more non-transitory computer-readable storage media of claim 13, wherein the instructions that, if executed, cause the computer system to monitor the execution, further cause the computer system to:

detect that the execution pipeline requires additional computing resources; and

allocate a second subset of the computing resource to the execution pipeline.

15. The one or more non-transitory computer-readable storage media of claim 14, wherein the second subset of the computing resource is allocated to a different execution pipeline.

16. The one or more non-transitory computer-readable storage media of claim 13, wherein the information associated with the configuration of the computing resource includes utilization associated with at least one of network bandwidth, CPU thread, local storage capacity, and local storage input/output bandwidth.

17. The one or more non-transitory computer-readable storage media of claim 13, wherein the instructions that, if executed, cause the computer system to monitor the execution, further cause the computer system to provide an outcome of the execution to a requestor.

18. The one or more non-transitory computer-readable storage media of claim 13, wherein the instructions that, if executed, cause the computer system to monitor the execution, further cause the computer system to provide an outcome of the execution as input data for a different execution pipeline.

19. The one or more non-transitory computer-readable storage media of claim 13, wherein the instructions, if executed, further cause the computer system to transmit, after executing the data processing request, the data as processed to a data storage system for storage.

20. The one or more non-transitory computer-readable storage media of claim 13, wherein the computing resource is a computational resource.