Selective parsing to enhance SIP performance

Abstract

The session initiation protocol (SIP) is used as the signaling protocol in the IP multimedia subsystem (IMS) and the signaling is becoming computing intensive comparing to the current telecommunication network. The SIP is a text-based protocol with characteristics of unordered and verbose headers, variable-size message, and case-insensitive keyword. It imposes challenges for an efficient message processing. The property of SIP elements being able to process SIP messages quickly is critical for the performance of IMS networks. This article investigates the performance of SIP message processed in SIP servers, mainly focusing on improving message parsing by introducing a method named selective parsing for SIP message (SP4SIP). By modeling and analyzing a SIP server with a tandem Jackson network, it is concluded that parsing messages is the bottleneck of a SIP server performance, i.e., it is the most processing intensive activity in the system. To validate the approach, it has been implemented in a high-performance SIP server in the authors' lab. The results show that selective parsing for SIP message can indeed reduce processing time.