Chapter 4: Resilience of WebSphere Portal Clusters Under Load

---

Overview

Executive summary: Recognizing the growing popularity and success of business-to-consumer (B2C) and business-to-employee (B2E) portals, the IBM High-Volume Web Site (HVWS) team studied failure and recovery scenarios in clustered, heavily loaded, WebSphere Portal systems, configured in topologies typical of IBM customer implementations.

We verified successful workload balancing when injecting a variety of different types of failure into the Web site infrastructure, and also when reenabling the unavailable system components. In studying workload management in these tests, our focus was on minimizing the effects on the portal's end users.

This chapter reviews our test methodology and provides several 'best practices' tips to help minimize the effects on portal end users of temporary failures within the Web site infrastructure.

We obtained the recommendations in this chapter by studying WebSphere Portal as a particular WebSphere application; many of the recommendations may apply equally to clustered WebSphere applications in general.

---