SERVER CONSOLIDATION SYSTEM

• US 20120101968A1
• Filed: 10/22/2010
• Published: 04/26/2012
• Est. Priority Date: 10/22/2010
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Banerjee, Pradipta K., Mylavarapu, Swarna L., Sukthankar, Vijay
• Granted Patent: US 10,797,953 B2
• Time in Patent Office: Days
• US Class Current: 706/21
• CPC Class Codes: H04L 41/0823 characterised by the purposes of a change of settings, e.g. optimising configuration for enhancing reliability for optimising operational conditions of wireless networks H04W24/02
• CPC Class Codes: H04L 41/0893 Assignment of logical groups to network elements

Abstract

A computer program product for a network management device, including: a computer readable storage medium to store a computer readable program, wherein the computer readable program, when executed on a computer, causes the computer to perform operations for server management in a computer network. The operations include: receiving resource usage data generated at a network communication device coupled between a server and a network management device, wherein the resource usage data describes resource usage of the server; and classifying the server into a cluster of servers based on the resource usage data from the network communication device and a cluster characterization for the cluster. The cluster includes a plurality of servers with similar resource usage data, and the cluster is one of a plurality of clusters managed by the network management device.

Claims

1. A computer program product for a network management device, comprising:

a computer readable storage medium to store a computer readable program, wherein the computer readable program, when executed on a computer, causes the computer to perform operations for server management in a computer network, the operations comprising;

receiving resource usage data generated at a network communication device coupled between a server and a network management device, wherein the resource usage data describes resource usage of the server; and

classifying the server into a cluster of servers based on the resource usage data from the network communication device and a cluster characterization for the cluster,

wherein the cluster includes a plurality of servers with similar resource usage data,

wherein the cluster is one of a plurality of clusters managed by the network management device.

2. The computer program product of claim 1, wherein the resource usage data comprises only network usage of the server generated at the network communication device.

3. The computer program product of claim 2, wherein the computer program product, when executed on the computer, causes the computer to perform additional operations, comprising:

predicting processor usage and memory usage for the server based on the network usage.

4. The computer program product of claim 1, wherein the computer program product, when executed on the computer, causes the computer to perform additional operations, comprising:

updating the cluster characterization for the cluster after the server is classified.

5. The computer program product of claim 1, wherein classifying the server into the cluster further comprises:

training a neural network model to correctly identify clusters based on the resource usage data for all of the servers in each cluster; and

using the neural network model to correctly classify the server into the cluster and to predict additional resource usage data for the server.

6. The computer program product of claim 1, wherein the computer program product, when executed on the computer, causes the computer to perform additional operations, comprising:

calculating a server characterization of the resource usage data for the server for comparing to the cluster characterization when classifying the server.

7. The computer program product of claim 6, wherein the server characterization comprises at least one value of the group consisting of mean, variance, 50^th percentile, 60^th percentile, 70^th percentile, 80^th percentile, and 90^th percentile.

8. A method for server consolidation in a computer network, comprising:

receiving resource usage data generated at a network communication device coupled between a server and a network management device, wherein the resource usage data describes resource usage of the server; and

classifying the server into a cluster of servers based on the resource usage data from the network communication device and a cluster characterization for the cluster,

wherein the cluster includes a plurality of servers with similar resource usage data,

wherein the cluster is one of a plurality of clusters managed by the network management device.

9. The method of claim 8, wherein the resource usage data comprises only network usage of the server generated at the network communication device.

10. The method of claim 9, wherein the computer program product, when executed on the computer, causes the computer to perform additional operations, comprising:

predicting processor usage and memory usage for the server based on the network usage.

11. The method of claim 8, wherein the computer program product, when executed on the computer, causes the computer to perform additional operations, comprising:

updating the cluster characterization for the cluster after the server is classified.

12. The method of claim 8, wherein classifying the server into the cluster further comprises:

training a neural network model to correctly identify clusters based on the resource usage data for all of the servers in each cluster; and

using the neural network model to correctly classify the server into the cluster and to predict additional resource usage data for the server.

13. The method of claim 8, wherein the computer program product, when executed on the computer, causes the computer to perform additional operations, comprising:

calculating a server characterization of the resource usage data for the server for comparing to the cluster characterization when classifying the server,

wherein the server characterization comprises at least one value of the group consisting of mean, variance, 50^th percentile, 60^th percentile, 70^th percentile, 80^th percentile, and 90^th percentile.

14. A network management device, comprising:

a processing device configured to;

receiving resource usage data generated at a network communication device coupled between a server and the network management device, wherein the resource usage data describes resource usage of the server; and

classifying the server into a cluster of servers based on the resource usage data from the network communication device and a cluster characterization for the cluster,

wherein the cluster includes a plurality of servers with similar resource usage data,

wherein the cluster is one of a plurality of clusters managed by the network management device.

15. The device of claim 14, wherein the resource usage data comprises only network usage of the server generated at the network communication device.

16. The device of claim 14, wherein the computer program product, when executed on the computer, causes the computer to perform additional operations, comprising:

predicting processor usage and memory usage for the server based on the network usage;

and updating the cluster characterization for the cluster after the server is classified.

17. The device of claim 14, wherein classifying the server into the cluster further comprises:

training a neural network model to correctly identify clusters based on the resource usage data for all of the servers in each cluster; and

using the neural network model to correctly classify the server into the cluster and to predict additional resource usage data for the server.

18. The device of claim 14, wherein the computer program product, when executed on the computer, causes the computer to perform additional operations, comprising:

calculating a server characterization of the resource usage data for the server for comparing to the cluster characterization when classifying the server,

wherein the server characterization comprises at least one value of the group consisting of mean, variance, 50^th percentile, 60^th percentile, 70^th percentile, 80^th percentile, and 90^th percentile.

19. A network management system, comprising:

a network management device having a processing device; and

a network communication device in communication with the network management device,

wherein the processing device is configured to;

receive resource usage data for the servers;

group the servers into clusters based on a server characterization of the resource usage data for each server;

collect network usage data for a new server from the network communication device;

classify the new server into a cluster based on the network usage data for the new server and a cluster characterization for the servers in the cluster,

wherein servers in the cluster have similar network usage characteristics; and

update the cluster characterization for the cluster after the new server is classified.

20. A computer program product for a network management device, comprising:

a computer readable storage medium to store a computer readable program, wherein the computer readable program, when executed on a computer, causes the computer to perform operations for server consolidation in a computer network, the operations comprising;

receiving resource usage data for a plurality of servers;

grouping the servers into clusters based on a server characterization of the resource usage data for each server; and

classifying a new server into a cluster based on a subset of resource usage data for the new server and a cluster characterization for the servers in the cluster,

wherein servers in the cluster have similar characteristics for the subset of resource usage data.

21. The computer program product of claim 20, wherein the subset of resource usage data comprises only network usage of the new server obtained from a network communication device in communication with the network management device, wherein the resource usage data for the plurality of servers comprises processor usage, memory usage, and network usage for each server.

22. The computer program product of claim 20, wherein the subset of resource usage data is less than all of the resource usage data used for grouping the plurality of servers into clusters.

23. The computer program product of claim 20, wherein the computer program product, when executed on the computer, causes the computer to perform additional operations, comprising:

calculating upper and lower fence values for the server characterization; and

eliminating outlier values from the server characterization that fall above the upper fence value or below the lower fence value.

24. The computer program product of claim 20, wherein the computer program product, when executed on the computer, causes the computer to perform additional operations, comprising:

updating the cluster characterization for the cluster after the new server is added.

25. The computer program product of claim 20, wherein classifying the new server into the cluster further comprises:

training a neural network model to correctly identify clusters based on the subset of resource usage data for all of the servers in each cluster; and

using the neural network to correctly classify the new server into the cluster and to predict additional resource usage data for the new server.