Dynamic cluster is the new Intelligent Management feature, which enables WebSphere 8.5 to dynamically stop, start and allocate resource as and when, required for an application running on a dynamic cluster. Dynamic cluster works in two modes: Add operation & Remove operation.

Add operation- in Add operation, it add resources when all servers in a dynamic cluster are started and demand is still not satisfied. This is possible through automatic actions, which you have defined under the Add operation.

Remove operation- in Remove operation, it stops the servers when demand is very minimal or if running resources are not required to serve the requests. This operation will never shutdown below the number of minimum active server instance configured in Remove operation.

Operational Mode

Operation Mode is the method that interacts with the dynamic operation environment. This has three modes: Manual, Supervised & Automatic.

Manual- it’s similar to the normal application server environment with static cluster. Manual operation mode is as good as WebSphere application server traditional cluster.

Supervised- provides information and recommendation for actions that WebSphere administrator can approve or deny.

Automatic- the intelligent management takes action automatically. Dynamic operations environment handles add/remove operation and provisioning.

Creating Dynamic Cluster

Go to Servers >> Clusters >> Dynamic clusters

Select Operational mode as Automatic and click on New

Select Server type as WebSphere application server and click on Next

Leave default method – Automatically define cluster members with rules

Provide Dynamic cluster name – First Dynamic Cluster and click on Next

You can select multiple Node Group if you have, in this course we have only one Node group as Default Group. Leave default setting and click on Next

Select dynamic cluster template as default and click on Next

Specify dynamic cluster specific properties are the one you configure how dynamic cluster should behave. We will leave the default settings as following and click on Next.

Keep one instance started at all times – dynamic cluster will be always active with one instance. By configuring this we ensure there is no application outage.

Do not limit thee number of instances that can start- dynamic cluster will start all instances in defined node as demand increases. By configuring this we ensure all resources are utilized when demand is high.

No isolation requirements- as we have only one Node Group we can continue with no isolation requirements.

Review the configuration summary and click on Finish to create Dynamic Cluster

Save the configuration and synchronize changes with nodes. Once done, you can see newly created Dynamic cluster is listed.

You can also see dynamic cluster member is created under Application servers and started one instance as we have selected to keep one instance started all times.

Now in this dynamic cluster setup, one instance will be always running. However, as demand increase other cluster instance also will be started.

Learn Health Management in IBM WebSphere Application Server

Introduction

Health Management is part of WebSphere Virtual Enterprise environment, which is integrated in WebSphere Application Server 8.5. WebSphere 8.5 is integrated with Operational policies, which leverage Health Policies. Health Management is policy-driven approach to monitor the WebSphere enterprise application server usage and able to respond to the problem areas before outage occurs. Health Management has two elements: health controller and health policies.

Reaction Mode

Health Policies include the health condition, which you want to monitor in your environment. It reacts when your defined conditions are not met. There are two reaction modes: automatic or supervised mode.

Automatic mode: System will take an action when health policy violation is detected. For ex, if you configure to monitor memory usage and would like to restart JVM when message usage is 85% then system will restart targeted JVM when JVM heap size reach at 85%.

Supervised mode: System will create runtime task when health policy violation is detected. This requires manual intervention for WebSphere administrator to approve or deny the runtime task action.

Health Conditions

Health Condition is the object or metrics you want to monitor in your environment. There are eight following predefined health condition available in WebSphere 8.5. You do have an option to create custom health condition.

Age-based condition- this condition will monitor the defined JVM and take action when reaches a configured age threshold. Ex, you can configure this condition to restart JVM if it’s running for 15 days. Acceptable value for this condition is in Days or Hours as shown below.

Excessive request timeout condition- this condition will take action when the request timeout percentage exceeds the defined value. Acceptable value is in percentage as shown below.

Excessive response time condition- this will monitor the time it takes for request to complete and take action if the time exceeds the defined threshold. Ex, you can configure this condition to take thread dump when response time for request is one minute. Acceptable value is in Milliseconds, Seconds and Minutes as shown below.

Memory condition: excessive memory usage- monitors the memory usage of JVM and take action if it exceeds the threshold value. Ex, you can configure this condition to take JVM heap dump and restart JVM when memory usage exceeds the threshold. Acceptable value for JVM heap size is in percentage and offending time period in Seconds and Minutes as shown below.

Memory condition: memory leak- this will look for memory leaks on JVM and take action. This got three detection levels.

1. Fast (more false alarms)

2. Standard (some false alarms)

3. Slow (fewer false alarms)

Storm drain condition- monitor significant drop in the average response time and take action like generate thread dump and restart JVM. This got two Detection level.

1. Standard (some false alarms)

2. Slow (fewer false alarms)

Workload condition- this condition will detect once a JVM has served a configured number of requests. Ex, you can configure to restart JVM once it serves 20000000 requests.

Garbage collection percentage condition- this monitor percentage of time spent in garbage collection for defined time period and take action once exceeds the threshold. Acceptable value is percentage and sampling time period as shown below.

Health Action

Health Action is the health policy action to be run once exceeds the configured threshold. There are seven following predefined health action available in WebSphere 8.5. You do have an option to create custom health action.

Restart Server- to restart JVM
Take thread dumps– to take thread dumps of JVM
Take JVM heap dumps– to take JVM heap dumps
Generate an SNMP trap- generate SNMP trap for troubleshooting
Place server in maintenance mode- stop new client requests and serve only active session
Place server in maintenance mode and break affinity – stop new and existing action session
Place out of maintenance mode – ready to accept new requests

Health Policy Target

Health Policy or Action target can be JVM’s, Clusters, Dynamic clusters, on demand routers or Cells.

As next step, in next article I will talk about How to Create Health Policies

Free Resource: Download Optimizing Application Monitoring Free Whitepaper

websphere-infra

Monday, December 1, 2014

WebSphere 8.5.x Dynamic Cluster