Question

I'm looking for people who have had experiences scaling WebMethods ESB to large traffic volumes (both size and number of messages). How has that gone? Were there any issues and how did you solve them?

Was it helpful?

Solution

From the environments I've dealt with (from 4 to 1000 servers) it scales pretty well.

It depends wildly on the type of information transport technology you are managing.

Fastest is the proprietary webMethods Broker, which, on a well configured server, can easily handle millions of >100kb messages per day.

If you use a JMS transport directly on the Broker (no transform/repush on the native broker format) the extra message handling steps slows it a little bit (but the new 7.1.2 version has improved greatly).

Other types of transport (stateless web services and others) usually do not involve an ESB, but your logistics architecture may vary, so no clear answer there.


Most of the time, cloning the components in cluster or non-cluster systems is enough (the process is mostly IO-bound, so you might get good results on virtualization [or para-virtual, container, systems]; sometimes you really need more metal.

OTHER TIPS

AFAIK, people get some nice numbers with webMethods (but they do not use webMethods clustering). The webMethods flows can scale, with the downside that they are not persisted through each stage. If you don't use processes, you should be ok with the scaling.

Licensed under: CC-BY-SA with attribution
Not affiliated with StackOverflow
scroll top