Balancing workloads with clusters

 


+ Search Tips   |   Advanced Search

 

Overview

To monitor appservers and manage the workloads of servers, use server clusters and cluster members provided by the Network Deployment product.

To assist you in understanding how to configure and use clusters for workload management, below is a scenario. In this scenario, client requests are distributed among the cluster members on a single machine. (A client refers to any servlet, Java application, or other program or component that connects the end user and the appserver that is being accessed.) In more complex workload management scenarios, you can distribute cluster members to remote machines.

 

  1. Decide which appserver you want to cluster.

  2. Decide whether you want to configure replication domains and entries. Replication enables the sharing of data among processes and the backing up of failed processes. (Optional)

  3. Deploy the application onto the appserver.

  4. After configuring the appserver and the application components exactly as you want them to be, create a cluster. The original server instance becomes a cluster member that is administered through the cluster.

  5. If you did not do so when creating a cluster, create one or more cluster members of the cluster.

  6. Start all of the appservers by starting the cluster. Workload management automatically begins when you start the cluster members of the application server.

  7. Stop the cluster.

  8. Upgrade applications on clusters.

  9. Detect and handle problems with server clusters and their workloads.

  10. Tune the behavior of the workload management run time. (Optional)  

    If your application is experiencing problems with timeouts or your network experiences extreme latency, change the timeout interval for the com.ibm.CORBA.RequestTimeout property. Or,if the workload management state of the client is refreshing too soon or too late, change the interval for the com.ibm.websphere.wlm.unusable.interval property.

You need to define a bootstrap host for stand-alone Java clients, which are clients located on a different machine from the appserver that have no administrative server. Add the following line to the Java Virtual Machine arguments for the client: 

-Dcom.ibm.CORBA.BootstrapHost=machine_name

where machine_name is the name of the machine on which the administrative server is running.

Workload management (WLM)
Clusters
Replication
Workload management run-time exceptions
Clustering and workload management: Links