|
JavaOne: Filling a high availability gap
|
|
Tuesday, April 29, 2008
Reconfiguring server-side applications while they are up and running is tricky but absolutely essential for many high availability applications. If you are building an IPTV service, the service provider’s customers will not accept a reboot when the provider wants to modify a parameter.
Until a high availability standard becomes part of Java EE, developers can build a solution like the one which Kristiansson has specified together with Jens Jensen, an expert in open platform architectures at Ericsson. Their solution, which meets the strictest of high availability requirements, makes it relatively easy to deploy high availability applications. The solution is based on managed beans, which are part of the Java Management Extensions (JMX) API. Unlike other JMX solutions, Kristiansson’s and Jensen’s solution does not require the application to create the managed beans classes itself. The application only has to create an XML file that describes the structure of managed beans. This file is then read by a back-end, which creates the beans and makes them available to a management console. “What is unique about our solution is the back-end that we have built which allows the [managed] beans to be created dynamically,” Kristiansson says. “Of course, to use our solution, you will have to build the back end, so this is not a small scale hack.” Developers interested in the solution can find out more at Kristiansson’s Birds of a Feather session BOF-4945 – “Designing Manageable Java Platform, Enterprise Edition (Java EE platform) Applications in a Clustered Environment” – which is held on Thursday, May 8 at 7:30 pm. “It will be a mix of an architectural overview of our solution and a tutorial on how to use it,” Kristiansson says. By Olle Blomberg Last published April 29, 2008
|
|