You are on page 1of 2

Architecture Analysis for Performance and Scalability

Customer
The customer, a UK-based leading retail vendor, is in the business of selling garments to different classes of customersfrom employees of contract companies purchasing uniforms to generic customers purchasing garments, and so on. Their product was a portal catering to different types of customers.

Challenges
o Remedying frequent server crashes o Solving scalability issues when the peak load increased from 10 to 65 users o Improving peak load performance

Our Approach
o Analyze the current architecture and o Understand the application requirements o Review the architecture against stated business requirements and business challenges o Examine the application design and deployment environment o Realize the Performance & Scalability requirements by.. o Benchmarking the current performance requirements o Identifying the current performance statistics

o Fine-tune the current environment for optimum performance

o Test in production environment

o Devise an action plan for improvements

Our Solution
o ATG replaced the XSLT Parser with a patched version to solve the problem of frequent server crashes. DB Objects (Result Sets and Statements) were locked to address occasional server crashes. o Tweaked the JVM Heap in order to increase the peak load to 65 users. This was done by increasing the heap size to 2 GB, and using the Concurrent GC option. o Modified the design to reduce the session XML volumes to improve scalability. o Ported the application to WAS 5.1.1 from WAS4. o Suggested the use of plain old Java objects (POJOs) in place of XML for caching

o Recommended that XML be used only for data communication and representation

Customer benefits
o Fewer server crashes o The peak load increased from 10 users to 65 users o Concurrent user performance saw improvement

You might also like