For some application I need a notion of 'shared state' within a cluster of computers within a network. The software will have to run on JBoss. Yesterday I had successfully setup a JGroups using the default configuration. Using a fine tutorial I had set up a shared state. JGroups provided a lot functionality so that I not have to work with sockets, multicast, discovery, etc. myself. However, transactions, 2 phase commits, serialization wasn’t standard functionality, which I needed too. I didn’t want to reinvent the wheel, so I decided to use JBoss Cache, which is built on top on JGroups.
I could successfully instantiate a cache, using the default factory class. However, a more proper way would be to have JBoss instantiate and manage the cache, since the application has to run on top of JBoss.
In the JBoss Cache documentation I’ve found the following entry:
CacheFactory factory = new DefaultCacheFactory(); // Build but don't start the cache // (although it would work OK if we started it) Cache cache = factory.createCache("cache-configuration.xml"); MBeanServer server = getMBeanServer(); // however you do it ObjectName on = new ObjectName("jboss.cache:service=Cache"); JmxRegistrationManager jmxManager = new JmxRegistrationManager(server, cache, on); jmxManager.registerAllMBeans(); ... use the cache ... on application shutdown jmxManager.unregisterAllMBeans(); cache.stop();
Sounds great doesn’t it? Well, first of all, it’s an awful lot of code just to get access to the cache. Secondly, there’s a reference to a configuration file you first have to place somewhere, and thirdly there’s the line that’s commented with:
// however you do it
Arg! Why can’t those people of JBoss just explain *how* I have to do that. You might say, a clustered cache is complicated stuff. Transactions, commits over network, multi casts. That’s all pretty complicated. However, this configuration crap has nothing to do with that. I *hate* configuration. I *hate* xml. I just want working stuff.
O well, most closed source software isn’t much better, if not much worse. Probably what the people behind JBoss wants users to do, is buy a support contract. Still a lot less expensive than software like Weblogic or Websphere.
Last year I worked with a collegue on a software project. As I architect I set up a backend. As frontend technology we used flex. As backend technology, we used Spring, JPA (via Toplink) and Java. Communication between frontend and backend was accomplished by using Blazeds.
The stack was very productive, and after a short learning curve, we could implement software quickly.
A nice and informative article on how to use these technologies is recently written by the collegue I worked with. The article is in Dutch, for those that can read that language see: Stack Attack: Flex, Spring, JPA en Toplink
subscribe via RSS