This content has been marked as final. Show 2 replies
user738616 wrote:Given the different subnets, then using WKA and Unicast is probably preferred over Multicast with a higher TTL setting.
I am trying to use Coherence*Web for HTTP Session Management and the details are as under:
Environment: Coherence Grid is runing in a different subnet than the App Servers
Please help me with the following questions:
1. What would be the ideal architecture for HTTP Session Management in the given environment?
2. What would be the performance implication of using Coherence*Extend over the Distributed Caching? Any performance test results available?Since there are extra network hops involved, the performance impact is non-zero. Plenty of customers have used an Extend configuration, the impact on your application will depend on your network performance and application session size. We have not published any performance numbers.
3. What could be the issues if Extend is used for HTTP Session Management?Extend is supported with Coherence*Web, please see the product documentation for the special configuration recommendations.
Thank you and look forward to hearing from you.
Thanks for the reply!
Are you recommending to use Distributed Cache (WKA and Unicast with a higher TTL setting) over Coherence*Extend in this environment (Coherence Grid on a different subnet than Application Server)?
Based on my understanding using WKA and Unicast with a higher TTL setting for this environment might lead to split brain scenario and also will loose the flexibility of growing the number of Coherence nodes by the speed of WAN. Please correct me if I may be wrong.
Also, what would be the recommended topology - near backed by distributed or distributed (Coherence*Web)?