Pauseless Garbage Collection Feature Story

Are small memory heaps limiting your application’s growth and functionality? Are your response times suffering from long garbage collection pauses? Determining the right memory heap size in object oriented applications is an incredibly complex and unforgiving affair. Make your heap a little too big and your service levels are blown with painfully long garbage collection pauses. Make it slightly too small and your application crashes with out-of-memory errors. The net result: applications are all too often constrained to fit in unnaturally small heap sizes, severely restricting the application’s ability to grow or deliver enhanced functionality and putting unnecessary pressure on the database that is expensive to relieve!

Multinational Bank Doubles Transaction Volume and Eliminates Crippling Bottlenecks

Most Java™ platform based applications grapple with heap limitations. A large bank’s technical staff had recently spent considerable time and effort optimizing heap sizes, but without the right platform, only so much could be done and some important business requirements were left exposed:

• End-user response time. Traders were experiencing slow and irregular response time due to excessive pause times on the order of 10 seconds. The direct result was missed trades and lost revenue opportunities for the bank.

• Application functionality. In an effort to minimize heap size, IT staff disabled some of the application’s most compute intensive features. With limited access to key functionality such as risk analysis and ‘what if’ scenarios, traders made less informed trades and took on more risk.

• Geographic expansion. The bank wanted to roll out this capability in multiple locations around the globe. This required the application to simultaneously handle live trades from one location and end-of-day clearing activities from another location. The application couldn’t handle the increased load while meeting minimum service level requirements.

Traditional computing models can not address these issues without increasing cost and complexity elsewhere. The common prescription is to add more application servers, database servers, and additional layers of software. Why can’t the industry come up with a simpler lower cost solution? Well, it has! Read more »