Hi all,

We are still using Hazelcast for some common tasks even though we
alternatively have the RDBMS based coordination implementation. I am
planning to refactor code to remove those Hazelcast tasks in MB4.

I could see that we are using Hazelcast for following tasks,

   - Hazelcast node addresses are maintained in Node details for
   consumption by an MBean service. Since we are planning to use REST services
   as an alternative to admin service, we can get rid of this. Additionally,
   in a non-Hazelcast environment, it does not make sense to maintain the
   Hazelcast node addresses.
   - To check if the node IDs are duplicated, a Hazelcase map is
   maintained. We can remove this when we use the RDBMS based coordination.
   - A unique ID generated using a Hazelcast ID generator is used as a
   parameter in generating cluster-wide unique IDs. Since we are focusing
   mainly on the 2 node HA deployment we do not need a cluster-wide unique
   message ID. If we need a cluster-wide unique ID we can later implement an
   algorithm on top of the RDBMS based coordination.


WDYT?

-- 
Asanka Abeyweera
Senior Software Engineer
WSO2 Inc.

Phone: +94 712228648
Blog: a5anka.github.io

<https://wso2.com/signature>
_______________________________________________
Dev mailing list
[email protected]
http://wso2.org/cgi-bin/mailman/listinfo/dev

Reply via email to