Increasing performance of a streaming application by running experimental permutations

• US 9,237,079 B2
• Filed: 11/19/2013
• Issued: 01/12/2016
• Est. Priority Date: 11/19/2013
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Bragstad, Lance, Branson, Michael J., Cao, Bin, Carey, James E., Odden, Mathew R.
• Primary Examiner(s): Zhang, Shirley
• Application Number:
  US14/084,168
• Publication Number:
  US 20150142951A1
• Time in Patent Office: 784 Days
• US Class Current: 1/1
• CPC Class Codes: G06F 9/455 Emulation; Interpretation; Software simulation, e.g. virtualisation or emulation of application or operating system execution engines
• CPC Class Codes: G06F 9/45533 Hypervisors; Virtual machine monitors
• CPC Class Codes: G06F 9/50 Allocation of resources, e.g. of the central processing unit [CPU]
• CPC Class Codes: G06F 9/505 considering the load
• CPC Class Codes: H04L 41/5019 Ensuring fulfilment of SLA
• CPC Class Codes: H04L 43/0817 by checking functioning
• CPC Class Codes: H04L 67/1001 for accessing one among a plurality of replicated servers

Abstract

A streams manager monitors performance of a streaming application, and when the performance needs to be improved, the streams manager automatically requests virtual machines from a cloud manager. In response, the cloud manager provisions one or more virtual machines in a cloud. The streams manager builds a permutation of a flow graph for the streaming application using the virtual machine(s), runs the permutation in parallel with the corresponding portion of the flow graph, and compares the performance of the original portion of the flow graph to the performance of the permutation. When the performance of the permutation exceeds the performance of the original part of the flow graph, the streams manager modifies the flow graph so the permutation replaces the original part of the flow graph. In this manner a streaming application can dynamically evolve to increase its performance as needed.

Claims

1. A computer-implemented method executed by at least one processor for improving performance of a streaming application, the method comprising:

executing a streaming application that comprises a flow graph that includes a plurality of operators that process a plurality of data tuples;

monitoring performance of the streaming application;

when performance of the streaming application needs to be improved, requesting a cloud manager to provision at least one virtual machine;

generating using the at least one virtual machine a permutation of a portion of the flow graph that provides logical equivalence of tuples of the portion of the flow graph;

running the permutation in parallel with the portion of the flow graph;

determining whether the performance of the permutation exceeds the performance of the portion of the flow graph; and

when the performance of the permutation exceeds the performance of the portion of the flow graph, the streams manager modifies the flow graph to replace the portion of the flow graph with the permutation.

2. The method of claim 1 wherein the permutation comprises a reordering of at least two operators in the flow graph.

3. The method of claim 1 wherein the permutation comprises a coalescing of at least two operators in the flow graph into one operator.

4. The method of claim 1 wherein the permutation comprises a split of at least one operator in the flow graph into multiple operators.

5. The method of claim 1 wherein, when the performance of the permutation does not exceed the performance of the portion of the flow graph by a defined threshold amount, the streams manager does not modify the flow graph to replace the portion of the flow graph with the permutation.

6. The method of claim 5 wherein, when the performance of the permutation does not exceed the performance of the portion of the flow graph by the defined threshold amount, the streams manager identifies a second permutation to the flow graph and generates the second permutation to determine whether the second permutation improves the performance of the streaming application.

7. The method of claim 1 wherein the streams manager determines when the performance of the streaming application needs to be improved by comparing current performance of the streaming application to at least one performance threshold.

8. A computer-implemented method executed by at least one processor for improving performance of a streaming application, the method comprising:

executing a streaming application that comprises a flow graph that includes a plurality of operators that process a plurality of data tuples;

monitoring performance of the streaming application;

when performance of the streaming application needs to be improved, requesting a cloud manager to provision at least one virtual machine, wherein the streams manager determines when the performance of the streaming application needs to be improved by comparing current performance of the streaming application to at least one performance threshold;

generating using the at least one virtual machine a first permutation of a first portion of the flow graph that provides logical equivalence of tuples of the first portion of the flow graph, wherein the first permutation comprises a reordering of at least two operators in the flow graph;

generating using the at least one virtual machine a second permutation of a second portion of the flow graph that provides logical equivalence of tuples of the second portion of the flow graph, wherein the second permutation comprises a coalescing of at least two operators in the flow graph into one operator;

generating using the at least one virtual machine a third permutation of a third portion of the flow graph that provides logical equivalence of tuples of the third portion of the flow graph, wherein the third permutation comprises a split of at least one operator in the flow graph into multiple operators;

running the first, second and third permutations in parallel with the corresponding first, second and third portions of the flow graph;

determining whether the performance of each permutation exceeds the performance of the corresponding portion of the flow graph;

when the performance of a permutation exceeds the performance of the corresponding portion of the flow graph, the streams manager modifies the flow graph to replace the corresponding portion of the flow graph with the permutation;

when the performance of a permutation does not exceed the performance of the corresponding portion of the flow graph by a defined threshold amount, the streams manager does not modify the flow graph to replace the corresponding portion of the flow graph with the permutation.