Dynamic processing unit relocation in a multi-nodal environment based on incoming physical data

Patent metadata

Field	Value
Publication number	US-9686120-B2
Application number	US-91424910-A
Country	US
Kind code	B2
Filing date	Oct 28, 2010
Priority date	Oct 28, 2010
Publication date	Jun 20, 2017
Grant date	Jun 20, 2017

Abstract

A relocation mechanism in a multi-nodal computer environment dynamically routes processing units in a distributed computer system based on incoming physical data into the processing unit. The relocation mechanism makes an initial location decision to place a processing unit onto a node in the distributed computer system. The relocation mechanism monitors physical data flowing into a processing unit or node and dynamically relocates the processing unit to another type of node within the ‘cloud’ of nodes based on the type of physical data or pattern of data flowing into the processing unit. The relocation mechanism may use one or more rules with criteria for different data types observed in the data flow to optimize when to relocate the processing units.

First claim

The invention claimed is:

1. An apparatus comprising a computer system comprising a plurality of compute nodes each with a processor executing instructions in a memory to provide:
a first processing unit and a second processing unit stored in the memory of the compute nodes and wherein the first processing unit and the second processing unit are part of an application executed by the processors of the compute nodes;
a relocation mechanism that monitors data flowing from the first processing unit into the second processing unit to determine a type of data in the data flow;
wherein the relocation mechanism determines to dynamically relocate the second processing unit to a different node of a preferred node type based on a relocation rule which includes criteria for the type of data found in the data flow into the second processing unit; and
wherein the relocation mechanism dynamically relocates the second processing unit by moving the second processing unit to an available node of the preferred node type, starting the second processing unit now located on the node of the preferred node type, and rerouting data from the first processing unit to the second processing unit.

2. The apparatus of claim 1 further comprising a rules file containing a plurality of relocation rules that defines criteria for relocating the processing units depending on different types of data in the data flow.

3. The apparatus of claim 1 wherein the relocation mechanism periodically samples data flowing into the processing units to determine the type of data.

4. The apparatus of claim 1 wherein the types of data determined by relocation mechanism include text data, floating point data, video data, audio data and encrypted data.

5. The apparatus of claim 1 further comprising a node configuration file that contains a preferred node list for a plurality of preferred node types.

6. The apparatus of claim 1 wherein the relocation mechanism chooses the different node for relocating the processing unit from the preferred list of nodes corresponding to the preferred node type.

7. An article of manufacture comprising software stored on a non-transitory computer readable storage medium, the software comprising:
a relocation mechanism that monitors a data flow from a first processing unit into a second processing unit to determine a type of data in the data flow wherein first processing unit and the second processing unit are part of an application executing on a plurality of nodes;
wherein the relocation mechanism determines to dynamically relocate the second processing unit to a different node of a preferred node type based on a relocation rule which includes criteria for the type of data found in the data flow into the second processing unit; and
wherein the relocation mechanism dynamically relocates the second processing unit by moving the second processing unit to an available node of the preferred node type, starting the second processing unit now located on the node of the preferred node type, and rerouting data from the first processing unit to the second processing unit.

8. The article of manufacture of claim 7 further comprising a rules file containing a plurality of relocation rules that defines criteria for relocating the processing units depending on different types of data in the data flow.

9. The article of manufacture of claim 8 wherein the relocation mechanism periodically samples data flowing into the processing units to determine the type of data.

10. The article of manufacture of claim 8 wherein the types of data determined by relocation mechanism include text data, floating point data, video data, audio data and encrypted data.

11. The article of manufacture of claim 8 further comprising a node configuration file that contains a preferred node list for a plurality of preferred node types.

12. The article of manufacture of claim 8 wherein the relocation mechanism chooses the different node for relocating the processing unit from the preferred list of nodes corresponding to the preferred node type.

Assignees

• Barsness Eric L
• Branson Michael J
• Santosuosso John M
• IBM

Inventors

• Barsness Eric L
• Branson Michael J
• Santosuosso John M

Classifications

• G06F9/5033 Primary

  considering data affinity · CPC title

• H04L29/08144 Primary

  Electricity · mapped topic

• H04L67/1002

  Electricity · mapped topic

• G06F9/5044

  considering hardware capabilities · CPC title

• H04L67/1001 Primary

  for accessing one among a plurality of replicated servers · CPC title