Parallel SIP Proxy Servers Using Direct Routing Approach

Abstract

In this paper we propose direct routing approach to SIP proxy load balancing. Direct routing in combination with failover techniques, also described in this paper, is probably the best choice for introduction of a redundant SIP proxy servers. Our approach is based on a standard IPVS (IP virtual server) modules distributed within GNU/Linux kernel. IPVS is additionally extended with our SIP scheduler, which is necessary for SIP traffic distribution across server farm. Failover solution is based on VRRP (virtual router redundancy protocol), and it is not in depth analyzed in this paper. Proof of quality of direct routing concept is presented through series of measurement results. Measurements are conducted on a test network and by the principals of the Sip Stone proposal. The measurement result show that system capacity is proportional to the number and capacity of each server added to the SIP proxy farm