Turbo SIP Performance: WebLogic SIP Server on Intel and JRockit
Ken Lee's Blog |
September 28, 2005 12:17 AM
|
Comments (3)
BEA, in cooperation with Intel, have produced SIP message and call processing performance benchmarks which have blown-away industry norms for busy hour call attempts (BHCA), and message processing latency.
Using BEA's SIP application server, BEA WebLogic SIP Server, and running it on Intel Xeon blade servers running Red Hat Linux on the BEA JRockit JVM, BEA and Intel have posted SIP performance benchmarks of well over 10M BHCA, which translates to 500 calls per second, while retaining low SIP message processing latency of under 40msec. This compares with industry norms of handling 1M BHCA per node, while processing messages in less than 200msec.
As developers and telecom network operators deploy SIP and IMS based new services, it is critical that the SIP and IMS network architects take into consideration the performance issues as more and more complex, converged IT-telecom applications are brought to market. And as operators plan out an IMS-based network architecture, it's critical for them to size their respective markets separately, and be able to do effective subscriber as well as network capacity planning. Larger the capacity requirements, the more important the WebLogic SIP Server-like performance numbers become more attractive, and in fact, very business critical to network operators.
Comments
Comments are listed in date ascending order (oldest first) | Post Comment
-
Impressive stuff! Out of interest, what defines a "call" in these statistics? Are you basing this on SIPStone, or some other benchmark for "call" in the SIP world? Is this for proxying or B2BUA call scenarios?
It'd be great if you could explain more about what these figures are specifying; without the extra info they don't mean very much at all... ;-)
Posted by: hepwori on September 28, 2005 at 9:46 AM
-
No, this benchmark testing with Intel is not based on the Columbia-Ubiquity SIPstone benchmark. Use case is a SIP proxy servlet, with each call consisting of an average 5 SIP messages.
The call set-up latency time is measured in milliseconds from the time the SIP proxy servlet receives an INVITE from the User Agent Client to the time when the ACK is forwarded to the User Agent Server.
- Ken
Posted by: jusmagk on September 28, 2005 at 10:32 AM
-
Thanks for the extra info; useful stuff. Can you share any details about the hardware behind these performance figures? Number of blades, processors, GHz etc?
Posted by: hepwori on September 28, 2005 at 1:08 PM
|
