Implementing Clustering in GlassFish Version 2

This article discusses the cluster features of GlassFish Version2 to help you deploy your own applications on the GlassFish cluster.

Sun Java System Application Server 9.1 is the Sun release of the Open source GlassFish version 2 application server. This article uses the name GlassFish version 2 to denote both.

Basic concepts

Clusters in the application server enhance scalability and availability, and the two are interrelated.

To provide highly available services, software systems must have the following features:

The system must be able to create and run multiple instances of the service-provided entity. In an application server, a service-provided entity is a Java EE application Server instance running in a cluster, and the service is a Java EE application that is deployed.

To handle the increased service load, the system must be able to add application server instances to the cluster to accommodate larger deployments.

If a single application server instance in the cluster fails, you must be able to failover to another server instance so that the service is not interrupted. Although the failure of a server instance or a physical machine can degrade the overall quality of the service, it is not acceptable to complete a service outage in a high-availability environment.

If a process wants to modify the state of a user's session, the session state must be persisted and can be recovered after the process restarts. The easiest way to do this is to maintain a reliable copy of the session state, and if the process terminates, you can resume session state when the process restarts. This principle is similar to the principle of a highly available RAID storage system.

These requirements result in the system sacrificing efficiency for high availability.

To support scalability and high availability, the GlassFish Application Server provides the following server-side entities:

Server instance (server Instance) – The servers instance is a Java EE server process (GlassFish Application server) that runs various Java EE applications. Depending on the requirements of the Java EE specification, each server instance must be configured for each subsystem that is to be run.

Node Agent – The node agent is a proxy process that runs on each physical host that is running the server instance. When the Domain Administration Server (DAS), described later in this article, gives instructions, the node agent manages the lifecycle of the server instance.

Cluster (Cluster) – A cluster is a logical entity that determines the configuration of the server instance that makes up the cluster. The configuration of a cluster often means that all server instances in the cluster have the same configuration. Administrators typically view the cluster as a single entity and manage the server instances in the cluster using the GlassFish Admin console or the command-line interface (CLI).

You can create node proxies, server instances, and clusters when you install GlassFish, as described later in this article. Clusters and instances are organized into admin domains (administrative domain), which are controlled by Domain administration Server (DAS).

Domain Management architecture

The core concept of the GlassFish cluster architecture is the admin domain (administrative domains). Administrative domains represent access rights for administrators or Administrators groups. The following figure gives an overview of the Domain management architecture (contains only one domain).

Figure 1. Domain Management architecture