JMS vs. Hazelcast

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view

JMS vs. Hazelcast

Jason Cradit
Hello all,

I'm looking for recommendations on clustering 2+ Geoservers together.  I'll be building this in AWS and putting them behind a load balancer.  I have tools like EFS to create a shared data directory and RDS to build JDBC.  We really want this to be a pretty elastic environment to scale based on load.

So my use-case is one where I want all Geoservers to respond to the load balancer with the same data.  Like, if we change the Style of a layer on one Geoserver, all Geoservers know about it and respond with the update.  

I've built Hazelcast and it seems to work well.  My only issue is around the down operations.  Since we can't use multicast for joining, i'm building the tcp-ip field in the hazelcast.xml file dynamically at build time.  Works pretty well, but i'd be open to suggestions here.  Also, interested as it feels like it's not heavily maintained in the community.  Am I just out of touch here?  It seems like Boundless was the primary driver behind this, historically.

I haven't built JMS, but reading the documentation it looks pretty complex for what we are trying to do I'm concerned about resiliency in an elastic environment and any guidance for building it in AWS with AmazonMQ.  This feels more of the mainstream clustering option for Geoserver from what i've read.

I appreciate all your input and guidance.


Geoserver-users mailing list

Please make sure you read the following two resources before posting to this list:
- Earning your support instead of buying it, but Ian Turton:
- The GeoServer user list posting guidelines:

If you want to request a feature or an improvement, also see this:

[hidden email]