问题
I have a Scenario where i have 2 weblogic servers let say WL1 and WL2 in WL1 i have 2 applications deployed APP1 and APP2 in WL2 i have 2 applications deployed APP3 and APP4 I want to create a infinispan configuration where APP1 from WL1 forms a cluster with APP3 in WL2 and APP2 from WL1 forms a cluster with APP4 in WL2
So i tried using default UDP multicasting and looks like all 4 applications are forming a cluster, so i changed the multicast port to solve this issue but is this the only way to get across this kind of a situation?
Can something be done with TCPPing i am wondering because it's a p2p so it can form a cluster between WL1 & WL2 and not with individual applications right?
I am also considering of using remote caching but want's to explore embedded caching before we completely rule it out, so any help will be highly appreciated.
回答1:
Answering to the question in comment: remote x embedded
The main drawback of remote cache is the latency added for communication between the client and server. Also, you can't use transactions in remote mode, there may be other features missing as well. On the other hand, with remote cache you can easily upgrade the application without changing the data inside Infinispan. With embedded mode this would be more complicated. You can also load-balance: although Infinispan is aiming at linear scalability, things are never as bright. Therefore, you can use e.g. 20 application servers and only 4 Infinispan servers (provided that the application requires more computing power).
来源：https://stackoverflow.com/questions/19670244/infinispan-clustering-applications-on-2-servers