Next generation clustered JEE applications will run in a Grid environment – essentially, the deployed applications will tap into the resources of the Grid to scale on demand dynamically. At peak loads, the deployment could scale to hundreds of Grid nodes. The scalability characteristics of applications often are limited by how quickly or well the data can be shared across such a Grid environment.
With continuously changing conditions, you may have to share object state across hundreds of servers, or, bombard one or more centralized disk databases with thousands of concurrent requests. Very quickly the Grid nodes, predominantly meant to maximize CPU utilization, are waiting for either data to arrive or routed to some external data repository.
The talk introduces a new class of data management software – a highly distributed, main memory based infrastrucuture that can automatically sense the demand and respond by auto provisioning data across many nodes in the Grid. The result is that now the clustered JEE application is able to do highly parallelized data access, where the data is efficiently managed through a main memory storage Grid.
The focus will be on some of the concepts of distributed caching, techniques for data distribution ranging from optimistic to pessimistic models, garbage collection challenges and optimizations when dealing with data that is long lived and high in volume and ensuring continuous availability through multiple failure detection models.
Asif Shahid ‘s session on ‘
Grid enabling Data Intensive JEE applications’ was presented at the IndicThreads.com Conference On Java Technology 2007 held in Oct 2007 in Pune, India.
Presentation / Demo Code : Grid enabling Data Intensive JEE applications
* Video compressed for 256 kbps.