METHOD FOR INTELLIGENT PATCH SCHEDULING USING HISTORIC AVERAGES OF VIRTUAL I/O UTILIZATION AND PREDICTIVE MODELING

• US 20090187899A1
• Filed: 01/23/2008
• Published: 07/23/2009
• Est. Priority Date: 01/23/2008
• Status: Abandoned Application

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Rosas, Morgan J., Vallabhaneni, Vasu, Mani, Anbazhagan, Kovacs, Robert G.
• Application Number:
  US12/018,461
• Publication Number:
  US 20090187899A1
• Time in Patent Office: Days
• US Class Current: 717/168
• CPC Class Codes: G06F 8/65 Updates security arrangements therefor G06F21/57
• CPC Class Codes: G06F 9/4401 Bootstrapping security arrangements therefor G06F21/57

Abstract

A method for intelligent patch scheduling for a virtual (I/O) server is provided. Virtual I/O performance indicators of a virtual I/O server are monitored. The performance indicators are stored in a database. Historic averages of the performance indicators are maintained in the database. Patches to be applied to a client partition of the virtual I/O server are received. A reboot window is received for the client partition and is an allowed time frame for rebooting to apply the patches. Future virtual I/O utilization is predicted by running predictive modeling utilizing the historic averages of the performance indicators, and based on the predictive modeling, a specific time within the allowed time frame is determined for rebooting the client partition of the virtual I/O server to apply the patches. The virtual I/O server is rebooted to apply the patches to the client partition at the specific time within the reboot window.

Claims

1. A method for intelligent patch scheduling for a virtual input and output (I/O) server comprising:

monitoring virtual I/O performance indicators of a virtual I/O server;

storing the performance indicators in a database of the virtual I/O server;

maintaining historic averages of the performance indicators in the database;

receiving patches to be applied to a client partition of the virtual I/O server;

receiving a reboot window for the client partition, wherein the reboot window is an allowed time frame for rebooting the virtual I/O server to apply the patches;

predicting future virtual I/O utilization by running predictive modeling utilizing the historic averages of the performance indicators, wherein based on the predictive modeling, a module determines and selects a specific time within the allowed time frame for rebooting the client partition of the virtual I/O server to apply the patches;

wherein predictive modeling comprises;

selecting a utilization range from a plurality of utilization ranges for the virtual I/O server;

selecting a time window from a plurality of time windows for the virtual I/O server;

by utilizing the historic averages of the performance indicators, determining a probability that the virtual I/O server should be rebooted during the selected utilization range and determining a probability that the virtual I/O server should be rebooted during the selected time window;

to equal a total YES probability, combining the probability that the virtual I/O server should be rebooted during the selected utilization range with the probability that the virtual I/O server should be rebooted during the selected time window;

by utilizing the historic averages of the performance indicators, determining a probability that the virtual I/O server should not be rebooted during the selected utilization range and determining a probability that the virtual I/O server should not be rebooted during the selected time window;

to equal a total NO probability, combining the probability that the virtual I/O server should not be rebooted during the selected utilization range with the probability that the virtual I/O server should not be rebooted during the selected time window; and

comparing the total YES probability to the total NO probability; and

rebooting the client partition of the virtual I/O server to apply the patches at the specific time within the reboot window, responsive to the total YES probability being greater than the total NO probability.

2. The method of claim 1, wherein the reboot window comprises at least one of minutes, hours, days, and months.

3. The method of claim 1, wherein the virtual I/O performance indicators relate to a plurality of client partitions of the virtual I/O server;

and

wherein the virtual I/O performance indicators comprise at least one of a number of Internet protocol packets, a number of bytes transferred for virtual disk reads, a number of disk writes, and a number of files opened.