We are using Sonic MQ 6.1 CAA in our production environment since 2006 and as per my last post we are facing a frequency Directory Service unavailabilty issue since last few months.
As we are using both Java based JNDI client & ESB clients so this issue is creating problem for us now.
In our architecture we use single standalone DM, DS & agent Manager. DM , DS, AM is not fault tolerent with back-up.
We use 8 brokers with replication back-up for each broker and 4 clusters nodes presently for these brokers. All the queues & Topics are on these broker for message flow.
No# of esb clients connection to DM is 100+ and similarly for Java based clients is as per demand approx 500 +
we use custom services for transformation & different purpose example file read, write, ftp & all.
Server is SUSE linux VM, with 8 GB RAM. we are using two servers one is for hosting DM and other brokers while other is for hosting services. JVM config is min 256m & max 2048 for each configuration start-up.
Can any one help me in further tuning this infrastructure so that we can improve Directory Service Availabilty issue and can make it stabilized.
Moreover any input or business case for further improving SonicMQ Architecture or Upgrade is also welcome.
Presently I am planning to have a back-up Directory Service , but I am not very sure how it will improve stability. and I am also struck on the point what all changes I need to suggest to clients logic & code to have this back-up in use. As our ESB & JAVA based clients are geo distributed.
As I am also planning to purpose Business case to our client for its upgrade.