Users Guide

Introducing Clustering | Clustering
104 OMNM 6.5.3 Installation Guide
Introducing Clustering
Clustering transparently balances the computing load for this application’s EJB components—rule
engine, scheduler, logger, Business Object Manager (BOM), workspaces and mediation. This is
especially beneficial for the applications’ communication with the database storing its business
data.
By default, this application supports the distribution of its processes. It distributes the load per
client (not per request). Except for Mediation clustering functionality, fail-over or high availability
clustering is an optional add-on. To make a genuine highly available system, you must cluster
application servers, mediation servers, and database servers (Oracle RAC). Consult your sales
representative for the licensing requirements for fail-over, or high availability clustering.
The following are some of the benefits of clustering:
• Elimination of bottlenecks and single-point failures—Clustering several servers distributes
computing tasks, and enhances performance. You can even dynamically add a server to the
cluster to meet increasing user demand. Replication protects your application and users’ state
to ensure that the failures—like server crashes—can be fully masked from the user and
application.
• Transparency to your applications and application developers—developers do not have to deal
with intricacies of replication and load balancing. This means developers do not have to
modify their application components to run in a clustered environment.
• Hardware and OS independence—You can use clustering across disparate hardware and
operating system platforms.
OMNM Deployment Architecture
OpenManage Network Manager (OMNM) supports three primary deployment models:
Single-Server—The full application is installed on one server.
Distributed-Server—One or more servers are distributed.
Clustered/HA—multiple servers of each type may be used for performance gains or
to achieve High Availability.
The OMNM platform software architecture consists of the following principal run-time software
components:
Web Server—Eliminates the need for a separate Java client interface. Deployments
that have more that one web server or application server also require a load bal-
ancer.
Application Server—The system's central processing unit. It executes application
business logic. You can deploy it in both fault tolerant (Master/Slave) and cluster
configurations to limit downtime and optimize performance.
Upgrading application server first, if you are using the embedded database, also upgrades the
database, if necessary. It’s often easiest to install application server first simply because this
upgrade impacts any other application servers too, if they are clustered.
Database Server—Like the Application Server, you can deploy the Database Server in
a fault tolerant configuration to eliminate data loss during a system failure and to
ensure data integrity. This configuration typically uses Mysql replication or Ora-
cle RAC. You can cluster the Oracle database servers. See Installing Oracle on page
166. References to database servers below apply to all supported databases.