Clustering without J2EE App Server?

[jim_cross]

Is it possible to have a clustered cache without using an application server?

In particular, we're looking to have a set of Java services deployed to a number of regional servers. These services will be exposed as SOAP services, but will not sit within an app server (J2EE or otherwise).
The services will use Hibernate, and we're looking for a second level cache which will allow clustering between the caches.

Can JBoss Cache handle its own clustering, or does it require an application server to deal with the clustering?

Thanks,

Jim

[jim_cross]

...or alternatively, can Tomcat be used to provide clustering?

[brian.stansberry]

JBoss Cache can function and cluster just fine outside of an app server. An example of how to start one programatically:

TreeCache tree=new TreeCache();
PropertyConfigurator config=new PropertyConfigurator();
config.configure(tree, "replAsync-service.xml"); // read in config file from classpath
tree.createService();
tree.startService();

All caches on the network which share a common JGroups protocol stack definition in their config file will see each other and cluster.

[jim_cross]

Great, thanks Brian.

Also just found an example deployment in Tomcat :)

[brian.stansberry]

You can't use the cluster communication layer that comes with Tomcat to provide communication between JBoss Cache instances.

But, you can run JBoss Cache inside Tomcat; the code I just posted will work fine inside a web tier component running in Tomcat.