Clustering Samples
iPlanetTM Application Server Samples

Updated June 12, 2001

This document describes how to configure a simple cluster and exercise the cluster using sample applications. These exercises provide an introduction to the clustering support built into iPlanet Application Server.

1. Read the Sample Cluster Configuration Overview section.

2. Setup the Cluster

3. Exercise the Cluster Using the Sample Applications

HaServlet Sample : The HelloWorld of clustered applications.

HaJsp Sample: A simple clustered JSP sample.

HaSfsb Sample: A simple clustered stateful session bean.

Bank Sample: Servlets, JSPs and Stateless Session Beans

Java Pet Store Sample: JSPs, Session Beans and Entity Beans

Sample Cluster Configuration Overview

The cluster setup instructions to support these samples are meant produce an iPlanet Application Server configuration that is capable of demonstrating load balancing and session high availability. For simplicity, step-by-step instructions are included to set up a two machine cluster with web servers and app servers on each machine and a single directory server on one of the two machines. The directory server houses configuration data for both the application server as well as the business applications. The web server connectors and the application server instances consult the shared configuration data in the directory server to understand the location of application components as well as the processing characteristics (for example, load balancing algorithms, sticky vs. non-sticky, etc.) of both the application server and applications.

This simplistic cluster configuration is not representative of a true production configuration, but is sufficient to demonstrate the behavior of the application server. A production configuration would likely have the web servers on separate machines and a second directory server configured to act as a backup the first directory server.

Required for this exercise are two servers, either Windows or UNIX with the iPlanet Web Server installed. A heterogeneous cluster is possible, but instructions are not detailed here. This exercise could also be done on one UNIX box with two separate "ias homes" installed as two distinct users.

Configuration data is stored in the directory server on the first machine. When the second application server is installed, it will use the same directory server for configuration information. It is preferable to use the same tree in the directory server, allowing the same configuration information to be shared between all servers and web connectors. During installation of the second application server, you will enter the same value for cluster name and global config name as specified during the first application server install.

In this example, the web servers are on the same machine. If the web servers were housed on a separate tier of machines, then one would enter the same global config name and cluster name during the web connector install.

Copyright (c) 2001 Sun Microsystems, Inc. All rights reserved.