Designing the Compaq ProLiant Clusters HA/F100 and HA/F200
2-3
Compaq Confidential – Need to Know Required
Writer:
Bryan Hicks
Project:
Compaq ProLiant Clusters HA/F100 and HA/F200 Administrator Guide
Comments:
Part Number:
380362-003
File Name:
c-ch2 Designing the Compaq ProLiant Clusters HAF100 and HAF200.doc
Last Saved On:
8/24/00 12:00 PM
An active/active configuration has two primary designs:
■
The first design uses MSCS failover capabilities on both nodes, enabling
Node 1 to fail over clustered applications to Node 2 and enabling Node
2 to fail over clustered applications to Node 1. This design optimizes
availability since both nodes can fail over applications to each other.
■
The second design is a one-way failover. For example, the Microsoft
clustering software may be set up to allow Node 1 to fail over clustered
applications to Node 2, but not to allow Node 2 to fail over clustered
applications to Node 1. While this design increases availability, it does
not maximize availability since failover is configured on only one node.
When designing cluster nodes to fail over to each other, ensure that each
server has enough capacity, memory, and processor power to run all
applications (all applications running on the first node plus all clustered
applications running on the other node).
When designing your cluster so that only one node (Node 1) fails over to the
other (Node 2), ensure that Node 2 has enough capacity, memory, and CPU
power to execute not only its own applications, but to run the clustered
applications that can fail over from Node 1.
Another consideration when determining your servers’ hardware is
understanding your clustered applications’ required level of performance when
the cluster is in a
degraded state
(when one or more clustered applications is
running on a secondary node). If Node 2 is running near peak performance
when the cluster is in a normal operating state, and if several clustered
applications are failed over from Node 1, Node 2 will likely execute the
clustered applications more slowly than when they were executed on Node 1.
Some level of performance degradation may be acceptable. Determining how
much degradation is acceptable depends on the company.
Summary of Contents for ProLiant Clusters HA/F100
Page 8: ......
