Fuse ESB HA and DR

Latest response

Can anyone suggest the best approach for a Fuse Service Works architecture for High Availability (within a data center) and failover configuration to a DR data center? Just looking for some high-level architecture for Fuse clustering and multi-site configuration for failover.

Thanks!

Responses

Hi Michael,
The biggest bang-for-the-buck HA configurations are probably going to involve redundant load balancers at the front of a farm of servers offering stateless services. This configuration should be pretty resilient.
SwitchYard itself offers some clustering facilities, but these are largely centered around intra-service communications, and it rarely pays off to leave the machine you're processing on, so this probably shouldn't factor strongly into your design.
Use of Messaging (i.e. HornetQ) and the database should be given strong consideration. Hardening the database can be pretty straightforward. Trying to get failover out of Messaging can be a complicated effort, so allocate plenty of time for this if it's part of your plans.
Multi-site failover should be considered very carefully. The cost (in terms of performance trade-offs, etc) to maintain state across data centers can be high. If you can afford to write off the transactions in flight and accept a second data center that can immediately handle all 'new' traffic, this is a helpful concession. (Trying to fail-over transactions in flight with no loss of requests can be an expensive proposition. This will prove true of almost any software/product/etc. in use, it is no different for FSW).
Hope that helps,
Rick

Close

Welcome! Check out the Getting Started with Red Hat page for quick tours and guides for common tasks.