Scenario Overview

The objective of this customer scenario was to establish an environment in which an existing WebSphere® Application Server Network Deployment 6.1 cell is incorporated into a WebSphere® Application Server Network Deployment 7.0 cell with WebSphere Virtual Enterprise V6.1, sharing existing hardware. The basic premise of the scenario is to install WebSphere® Application Server Network Deployment 7.0 and Virtual Enterprise 6.1 on the same hardware of an existing WebSphere Network Deployment V6.1 cell, as a method of migrating applications to WebSphere Network Deployment 7.0 while still maintaining the existing WebSphere Network Deployment V6.1 cell.

This customer scenario was successfully implemented and validated in the IBM lab in March 2009. It was established based on environment requirements provided by an existing IBM customer. This solution to an environment requiring coexisting cells with different levels of Network Deployment is an IBM WebSphere Virtual Enterprise Best Practice recommendation. This solution is also valid with subsequent release levels of WebSphere Virtual Enterprise and WebSphere Network Deployment. Validation information can be found in the following validation section.

Audience

This document is divided into Parts 1 and 2. Part 1 is a general overview of this scenario, environment details, and is intended for technical sales, marketing, or the interested customer who wishes to learn more about the WebSphere products and how they can coexist.

Part 2 is more advanced and provides the technical details of implementing the scenario and is intended for administrators of WebSphere systems who will establish the new environment and perform the migration.

Part 1

Environment details

The configuration was established to identify a series of benchmarks and points of failure within a complex environment. The environment, which had scenarios to cover coexistence, was composed of the following items:

• WebSphere® Application Server Network Deployment 6.1.0.23 cluster (static cluster using default ports across four nodes)
• WebSphere® Application Server Network Deployment 7.0.0.3 cluster (dynamic cluster using non-default ports across four nodes)
• WebSphere® Virtual Enterprise 6.1.0.5 (Formerly Extended Deployment Operations Optimization)
• Two custom test applications

The environment was configured having one application (veApp1) installed in the WebSphere Application Server Network Deployment V6.1 environment and a second application (veApp2) installed into a dynamic cluster configured in the WebSphere Network Deployment 7.0.0.3 environment using Virtual Enterprise. Both veApp1 and veApp2 are the same application. A different name and context is used in order to identify test conditions between the different cells.

Both test applications were profiled equally with an equal level of load in order to establish peak operating characteristics for an extended period of time. The usage pattern applied to this scenario was a situation with the following desired characteristics:

• An environment having two critical high volume applications
• A requirement to stage an upgrade path while retaining the existing environment
• A requirement to leverage Virtual Enterprise in the new environment to manage resources from both environments

Implementation Diagram

The following diagram shows the relationship of the servers with regard to the physical nodes and the cells.

Application Placement Diagram

The diagram shows the relationship of the applications with regard to the physical nodes and the cells.

Validation information

The operational characteristics of the tests performed to verify the environment is as follows:

• Load driven from two JMeter controllers on separate comparable hardware. One for each application.
• Each script running on the Jmeter controller had two sections; one for each on demand router.
• Load targets were at 1000 virtual users for each controller with 500 assigned to each applcation passing through each On Demand Router.
• Ramp-up was set at 2000 seconds for each 500 virtual users. Approximately 4 virtual users started every 4 seconds.
• Once the test load leveled off (fully ramped), the On Demand Routers operated at approximately 90% - 95% CPU utilization.
• Each node operated at approximately 85% - 95% CPU utilization with several valleys during the run as the load cycled through operation.
• Each application established a session at the start of each cycle and closed them at the end of a cycle. Once the session was established, the application allocated a block of memory and performed and series of mathematical operations over a period of time before completing a cycle and invalidating the session.

Hardware Specifications

Part 2 Advanced setup