Load balancing and work distribution with Hazelcast

[opedge]

Hello. I'm developing high-scalable application, so I decided to use
Hazelcast for it. I have one frontend server, which puts messages for
nodes. Every node in cluster change it's workload in background thread
in distributed map, so, frontend server choose queue (every node has
it's own message queue) to put message in. My question is: Is
Hazelcast suitable for this design (we need workload distribution and
load balancing) or may be some alternatives? I like Hazelcast for it's
simplicity and nice design.

[Fuad Malikov]

Hi,

What do you mean by

" Every node in cluster change it's workload in background thread

> in distributed map".

And also could you explain your problem with more details if possible.
I will try to answer based on your explanations.

Thanks..

> --
> You received this message because you are subscribed to the Google Groups "Hazelcast" group.
> To post to this group, send email to haze...@googlegroups.com.
> To unsubscribe from this group, send email to hazelcast+...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/hazelcast?hl=en.
>
>

--

@fuadm

[opedge]

My system consists of 3 layers: frontend server (apache mina based,
which reads information from socket) and some application servers,
which process information from frontend server. Frontend server should
choose application server with minimum load to process request. So, I
think that I need something like jms, but it seems more complicated
than hazelcast.

[Talip Ozturk]

Frontend server can put all messages into "messages" distributed queue
and your application servers can consume this queue and process the
messages. If front-end server is expecting results of the processes
then you can use "results" distributed queue to pass results back to
frontend server. This way, you can dynamically add more application
servers to process more messages and application servers with more
CPUs can process more messages.

http://twitter.com/oztalip