Quote:
I am implementing a solution using hibernate search which has 11+ records to index in a clustered environment.
11 ?
Quote:
1) Remote Directory: As we plan to deploy our app on EC2 instances and mounting to remote directory to another EC2 is really tough or impossible. And S3 could be used but it looks this has performance issues.
really not recommended, file locking is a big problem on remote mounted directories.
Quote:
2) Infinispan directory: If I understand correctly, infinispan is in memory storage and with huge data it may hit the performance of application.
well actually we made it to be able to cluster better, it's definitely the best option for cloud deployments
Quote:
3) JMS/JGroups: This looks to be promising. But do I need any other JMS server we deploy it on tomcat?
yes, any JMS server will do. JGroups is an alternative to JMS, so using JGroups you won't need it.
Note that this doesn't solve the problem of where to *store* the index, in fact what we suggest is to combine the Infinispan directory with the JMS or JGroups approach. JGroups might seem simpler, but if the master crashes at least JMS would store the messages so it's considered more reliable.