can some one throw some light on how I can achieve high availability & data replication with 3 application server instances which belong to the same cluster ?
Am more interested in understanding how I should set up the SIP replication topologies among the three app server instances
This topic has been locked.
1 reply Latest Post - 2011-07-06T14:35:40Z by bpulito
Pinned topic SIP High Availability
Answered question This question has been answered.
Unanswered question This question has not been answered yet.
Updated on 2011-07-06T14:35:40Z at 2011-07-06T14:35:40Z by bpulito
bpulito 0600015YGA23 PostsACCEPTED ANSWER
Re: SIP High Availability2011-07-06T14:35:40Z in response to SystemAdminThe answer in this case is that you would have to have 3 servers in one of your replication domains if you had an odd number of app servers in your cluster. There is just no other way around this. We do support that configuration, it is just not ideal because you end up replicating more data than is needed to handle a failover. Note that you can NOT have the same app server reside in more than a single replication domain.
I think a better solution is to vertically scale your configuration by running multiple app server JVMs on each physical node. For instance, if every machine you deploy is running 2 application servers you would never have an issue where you would need more than two application servers per replication domain. Of course you want to make sure your replication domains are spanning physical machines.
One last option to mention here is that WXS is an alternate technology that can be used for SIP session replication (instead of DRS) and handles this case more cleanly because of the way backups are distributed across the cluster. Of course, WXS is an add on product and may not be acceptable due to the extra expense.