Browse Prior Art Database

Provisioning for Availability Management

IP.com Disclosure Number: IPCOM000126869D
Original Publication Date: 2005-Aug-04
Included in the Prior Art Database: 2005-Aug-04
Document File: 2 page(s) / 28K

Publishing Venue

IBM

Abstract

A method to to optimize the number of physical servers in different clusters for availability, defined as overall application uptime

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 57% of the total text.

Page 1 of 2

Provisioning for Availability Management

Run the following steps to optimize the number of physical servers in different clusters for availability.

1. Obtain the list of clusters (= logical servers) that are required for each business application. This information may already be available, for example, in IBM Tivoli Business Systems Manager*.

2. For each cluster, obtain host uptime statistics from the monitoring system for each host (= physical server) in the cluster. Uptime is the percentage of the time that a host functions as a server. Also, calculate an average uptime to be used as an assumed uptime for hosts that are added to that cluster.

3. Using the uptime formulas for hosts running in parallel in a cluster (cluster_uptime = 1-(1-host_1_uptime)*(1-host_2_uptime)*...*(1-host_n_uptime) ) and for an application that requires servers running in series (application_uptime = cluster_1_uptime * cluster_2_uptime * ... * cluster_n_uptime), calculate the overall uptime for the application. Notice that the formulas in this step use uptime as a fraction, between 0 (never available) and

1 (always avaliable).

4. Try and see if by moving hosts from the more reliable clusters to the less reliable clusters, you can obtain better uptime, by assuming the new hosts' uptime will be the average for that cluster.

For example,

1. An application requires IBM DB2* and IBM WebSphere Application Server*.

2. IBM DB2 in this configuration has an uptime of 80% (it's unreasonably low -...