Tuning B2B Server Engine Threads in SOA Suite 11g
By Shub Lahiri, A-Team on Apr 10, 2012
B2B 11g has a number of parameters that can be tweaked to tune the engine for handling high volumes of messages. These parameters are also known as B2B server properties and managed via the EM console.
This note highlights one aspect of the tuning exercise and describes the different threads, that can be configured to tune the performance of a B2B server.
The most common indicator of a B2B engine in need of a tuning is reflected in the constant build-up of messages in an internal JMS queue within the B2B server. It is called B2B_EVENT_QUEUE and can be monitored via the Weblogic server console. Whenever such a behaviour is seen, it usually results in general degradation of performance.
There could be many contributing factors behind a B2B server's degradation of performance. However, one of the first places to tune the server from the out-of-the-box, default configuration is to change the number of internal engine threads allocated within the B2B server.
Usually the default configuration for the B2B server engine threads is not suitable for high-volume of messaging loads. So, it is necessary to increase the counts for 3 types of such threads, by specifying the appropriate B2B server properties via the EM console, namely,
- Inbound - b2b.inboundThreadCount
- Outbound - b2b.outboundThreadCount
- Default - b2b.defaultThreadCount
The function of these threads are fairly self-explanatory. In other words, the inbound threads process the inbound messages that are coming into the B2B server from an external endpoint. Similarly, the outbound threads process the messages that are sent out from the B2B server. The default threads are responsible for certain B2B server-specific special tasks. In case the inbound and outbound thread counts are not specified, the default thread count also dictates the total number of inbound and outbound allocated threads.
As found in any tuning exercise, the optimisation of these threads is usually reached via an iterative process. The best working combination of the thread counts are directly related to the system infrastructure, traffic load and several other environmental factors.