Data Caching in a Clustered Environment

Question

          I want to cache read-only reference/code table data that will run in a clustered
          WLS6 environment. It's a JSP application and I am storing a complete HTML Select
          Control per reference/code table data in the cache. The question is where to
          cache it? I was going to put it in the ServletContext (JSP "application" implicit
          object), but the ServletContext is not replicated. I considered using JNDI, but
          there are problems with duplicate name errors when another server who doesn't
          originally bind the object tries to lookup, change and rebind the object. I guess
          JMS Multicasting is an option, but I don't want to implement JMS just for an application
          data cache.
          Any suggestions for a simple reference/code table read-only caching strategy that
          will work in a clustered WLS6 environment?
          

Answer 1

If the data is strictly read-only, and you do not have to worry about cache
          integrity, then look at WebLogic JSP cachetag:
          http://www.weblogic.com/docs51/classdocs/API_jsp.html#cachetag
          You can use it to cache both the output and the calculations results
          (variables calculated inside the cache tag).
          The scenario will be exactly the same for non-clustered and clustered
          cases - using multicast to broadcast small invalidation messages (so the
          data can be refreshed from the database) is ok, but replicating application
          data is not (and you definitely do not want to use JNDI for this purpose).
          BTW, the initial CacheTag implementation in 5.1 (supposedly) had a 'cluster'
          scope and I assume it was multicasting fresh data after cache miss - there
          is no such scope in 6.0 implementation.
          If you still want replication you can look at javagroups:
          http://sourceforge.net/projects/javagroups/
          (distributedhashtable example).
          Olsen <[email protected]> wrote:
          > Cameron,
          > Thanks for the reply. However, as I stated below, I am not interested in
          > JMS, nor an EJB solution to the problem. It really is not that complicated of
          > a concept and I know a solution or two (ServletContext, JNDI), but none that works
          > in a WLS6 clustered environment.
          > Any other ideas???
          > Thanks...
          > "Cameron Purdy" <[email protected]> wrote:
          >>Dimitri had a clever (as ever) solution using JMS to maintain cache
          >>integrity:
          >>
          >>explanation at
          >>http://dima.dhs.org/misc/readOnlyUpdates.html
          >>
          >>d/l from
          >>http://dima.dhs.org/misc/readOnlyUpdates.jar
          >>
          >>--
          >>Cameron Purdy
          >>Tangosol, Inc.
          >>http://www.tangosol.com
          >>+1.617.623.5782
          >>WebLogic Consulting Available
          >>
          >>
          >>"Olsen" <[email protected]> wrote in message
          >>news:[email protected]...
          >>>
          >>> I want to cache read-only reference/code table data that will run in
          >>a
          >>clustered
          >>> WLS6 environment. It's a JSP application and I am storing a complete
          >>HTML
          >>Select
          >>> Control per reference/code table data in the cache. The question is
          >>where
          >>to
          >>> cache it? I was going to put it in the ServletContext (JSP "application"
          >>implicit
          >>> object), but the ServletContext is not replicated. I considered using
          >>JNDI, but
          >>> there are problems with duplicate name errors when another server who
          >>doesn't
          >>> originally bind the object tries to lookup, change and rebind the object.
          >>I guess
          >>> JMS Multicasting is an option, but I don't want to implement JMS just
          >>for
          >>an application
          >>> data cache.
          >>> Any suggestions for a simple reference/code table read-only caching
          >>strategy that
          >>> will work in a clustered WLS6 environment?
          >>
          >>
          Dimitri