Optimizing performance in CEP systems via CPU affinity

• US 9,921,881 B2
• Filed: 06/27/2014
• Issued: 03/20/2018
• Est. Priority Date: 05/27/2014
• Status: Active Grant

• Current Assignee: Sybase Incorporated (SAP SE)
• Original Assignee: Sybase Incorporated (SAP SE)
• Inventors: Ahmed, Ibrahim, Gandhi, Palaniappan, Theiding, Mark
• Primary Examiner(s): Mobin, Hasanul
• Application Number:
  US14/318,279
• Publication Number:
  US 20150347509A1
• Time in Patent Office: 1,362 Days
• CPC Class Codes: G06F 9/5033 considering data affinity

Abstract

In an example embodiment, performance is optimized in a complex event stream (CEP) system. Information about a plurality of CEP threads is obtained. Then nearness among the plurality of CEP threads is determined, wherein nearness between a first and a second CEP thread indicates how much interaction is expected to occur between the first and second CEP thread. Based on the determined nearness, the plurality of CEP threads are organized into a plurality of CEP thread groups. Then, each of the plurality of CEP thread groups are assigned to a different processing node, with each processing node having one or more processors and a memory.

Claims

1. A method for optimizing performance in a complex event stream (CEP) system, the method comprising:

obtaining, using one or more hardware processors, information about a plurality of CEP threads;

determining, using the one or more hardware processors, nearness among the plurality of CEP threads, wherein determining nearness between a first and a second CEP thread indicates how much interaction is expected to occur between the first and second CEP thread and includes creating a Directed Acyclic Graph (DAG), with each of the plurality of CEP threads assigned a different node in the DAG and edges between nodes representing data dependencies among the threads, and wherein each edge in the DAG is assigned a nearness value, the nearness value between the first and second threads being measured by calculating an actual amount of data transmitted between the first and second threads during a set period;

organizing the plurality of CEP threads into a plurality of CEP thread groups, based on the determined nearness; and

assigning, using the one or more hardware processors, each of the plurality of CEP thread groups to a different processing node of a plurality of processing nodes, each processing node having one or more processors and a memory.

2. The method of claim 1, wherein each processing node is a non-uniform memory access (NUMA) node.

3. The method of claim 1, further comprising:

assigning each of the plurality of CEP threads a value representing processing capacity requested by a corresponding CEP thread; and

wherein the organizing is also based on the values for each of the plurality of CEP threads and information about processing capacity of each processing node.

4. The method of claim 1, wherein the method is performed during startup time.

5. The method of claim 1, wherein the method is performed during runtime.

6. The method of claim 1, wherein the nearness is determined at least partially based on one or more user hints, the one or more user hints indicating whether a particular thread is dynamic, semi-static, or static.

7. An apparatus comprising:

an event stream processing engine comprising;

an affinity analyzer executable by a processor and configured to;

obtain information about a plurality of CEP threads;

determine nearness among the plurality of CEP threads, wherein determining nearness between a first and a second CEP thread indicates how much interaction is expected to occur between the first and second CEP thread and includes creating a Directed Acyclic Graph (DAG), with each of the plurality of CEP threads assigned a different node in the DAG and edges between nodes representing data dependencies among the threads, and wherein each edge in the DAG is assigned a nearness value, the nearness value between the first and second threads being measured by calculating an actual amount of data transmitted between the first and second threads during a set period;

organize the plurality of CEP threads into a plurality of CEP thread groups, based on the determined nearness; and

assign each of the plurality of CEP thread groups to a different processing node, each processing node having one or more processors and a memory.

8. The apparatus of claim 7, wherein the apparatus is coupled to one or more input adapters receiving input from at least one of the following:

market data feeds, a message bus, network monitoring, application transactions, and internal event streams.

9. A non-transitory machine-readable storage medium comprising instructions, which when implemented by one or more machines, cause the one or more machines to perform operations for optimizing performance in a complex event stream (CEP) system, the operations comprising:

obtaining, using one or more hardware processors, information about a plurality of CEP threads;

determining, using the one or more hardware processors, nearness among the plurality of CEP threads, wherein determining nearness between a first and a second CEP thread indicates how much interaction is expected to occur between the first and second CEP thread and includes creating a Directed Acyclic Graph (DAG), with each of the plurality of CEP threads assigned a different node in the DAG and edges between nodes representing data dependencies among the threads, and wherein each edge in the DAG is assigned a nearness value, the nearness value between the first and second threads being measured by calculating an actual amount of data transmitted between the first and second threads during a set period;

organizing the plurality of CEP threads into a plurality of CEP thread groups, based on the determined nearness; and

assigning, using the one or more hardware processors, each of the plurality of CEP thread groups to a different processing node of a plurality of processing nodes, each processing node having one or more processors and a memory.

10. The non-transitory machine-readable storage medium of claim 9, wherein each processing node is a non-uniform memory access (NUMA) node.

11. The non-transitory machine-readable storage medium of claim 9, further comprising:

assigning each of the plurality of CEP threads a value representing processing capacity requested by a corresponding CEP thread; and

wherein the organizing is also based on the values for each of the plurality of CEP threads and information about processing capacity of each processing node.

12. The non-transitory machine-readable storage medium of claim 9, wherein the method is performed during startup time.

13. The non-transitory machine-readable storage medium of claim 9, wherein the method is performed during runtime.

14. The non-transitory machine-readable storage medium of claim 9, wherein the nearness is determined at least partially based on one or more user hints, the one or more user hints indicating whether a particular thread is dynamic, semi-static, or static.