Book

# Measuring SIP Proxy Server Performance

Authors:

## Abstract

Internet Protocol (IP) telephony is an alternative to the traditional Public Switched Telephone Networks (PSTN), and the Session Initiation Protocol (SIP) is quickly becoming a popular signaling protocol for VoIP-based applications. SIP is a peer-to-peer multimedia signaling protocol standardized by the Internet Engineering Task Force (IETF), and it plays a vital role in providing IP telephony services through its use of the SIP Proxy Server (SPS), a software application that provides call routing services by parsing and forwarding all the incoming SIP packets in an IP telephony network. SIP Proxy Server Performance closely examines key aspects to the efficient design and implementation of SIP proxy server architecture. Together, a strong design and optimal implementation can enable significant enhancements to the performance characteristics of SPS. Since SPS performance can be characterized by the transaction states of each SIP session, the book analyzes an existing M/M/1-network performance model for SIP proxy servers in light of key performance benchmarks, such as the average response time for processing the SIP calls and the average number of SIP calls in the system. It also presents several other real-world industrial case studies to aid in further optimizations. This book is intended for researchers, practitioners and professionals interested in optimizing SIP proxy server performance. Professionals working on other VoIP solutions will also find the book valuable. © 2013 Springer International Publishing Switzerland. All rights are reserved.

## Chapters (12)

During the 1990s, the telecommunication industries provided various PSTN services to the subscribers using an Intelligent Network (IN) called Signaling System Number 7 (SS7). SS7 was considered one of the most reliable and capable signaling protocols. IN is defined as an architectural concept for the operation and provision of services as recommended by ITU-T standard. Workstations installed with software interfaces with the network can provide many advanced services to the subscribers. Advanced Intelligent Network (AIN), the later version of IN, provided several telephone services like 800 service, call forwarding, call waiting, call pickup, follow me diversion, store locator service, and name delivery, which are valuable services to the traditional phone users. All around the world, telecommunication companies deployed the PSTN network for their subscribers to communicate from their homes or offices. Telephone subscribers are normally attached to a central office within their geographic area. These central offices have large-scale computers, generally called switches, built with huge hardware and managed by various software modules. These switches are considered as the brain of the PSTN network. Normally, there are several central offices within a metropolitan area. There are two widely used call types: (1) Local call where the called party is within the geographic area of the central office; (2) Long distance call or tandem call where the traffic outside the central office goes to one or more toll/tandem offices, which contains a tandem switch. To connect and send/receive traffic between the central offices and the tandem switches, trunks are being used. Routing the signaling messages with media streams between the switches over a packet-based network is the main function of SS7. Traditional telephone network has been deployed with advanced services to the telephone subscribers since the introduction of digital telecommunication switches. In the 1980s, the International Telegraph and Telephone Consultative Committee (CCITT) standardized the IN services in Q.1201. PSTN subscribers throughout the world are overwhelmed by the advanced services like call forwarding, call waiting, voice mail, speed dial, call return, and other simple user location services. ITU-U standard defines IN as a conceptual model and the “Framework for the design and description of the IN architecture”.
Both academia and telecommunication industries did several studies and published many literatures on the performance and reliability of PSTN based SS7 signaling. Authors in Wang (1991) studied the performance analysis of link sets on the CCS7 network based on SS7 signaling with complete derivation on the analytical queuing and delay models. Several performance models are based on SS7 traffic load matrix and traffic characteristics, network configuration and network protocols (Kohn et al. 1994). The same paper addresses the reliability for SS7 networks based on availability, dependability and robustness. In Chung et al. (1999), the team calculated the mean end-to-end delay of Common Channel Signaling Networks [CCSN] and used call arrival rate and failure rates. Total response time, network transmission time, and SCP processing are the key performance parameters considered by the work done in Atoui (1991).
At the beginning of our research, we identified that there are very limited research is being done on SIP proxy server performance and VoIP in general. In a very fast phase, due to the growth of Internet based technologies, IP telephony quickly became a direct replacement for PSTN. Lack of any SIP proxy server performance study is mainly because of the complexity involved in simulating or emulating the actual SIP proxy server performance models to accurately study the performance characteristics. We were motivated, excited and equally challenged to work in this area of research. We carefully evaluated our lab resources at CISCO “LASER” lab and leveraged that in our research to perform experiments with the actual proxy servers and collected our data.
During the course of this research, we realized that the performance of the proxy server, which is fundamentally a software entity, depends heavily on the software architecture of the system (that is, how the functionality is factored, and apportioned between various threads of execution, and various ways to communicate between functional modules). Decisions regarding this architecture are typically made by the software designers by taking into account the hardware and OS environment that is chosen for the realization, which in turn is driven by business decisions and other factors such as current support and expertise in a particular environment. In recent years, researchers have become aware that the most practically important proxy servers were more and more being based on a simple streamlined execution model, and also were extensively using multi-threading over predecessors. To make the performance study more realistic, it is a reasonable decision to educate on the software architecture of a real proxy server and collect experimental data on it as well. The earlier data set was collected from an open source proxy server implementation running on commodity hardware, as mentioned before. The basic understanding, which changed the modeling effort completely, is that such industry standard servers are increasingly using a single thread of execution to complete all processing needed by a single SIP request completely and sequentially, instead of using the older architecture of multiple concurrently running modules for the various parts of the functionality, coupled by message queues. To utilize the hardware well, concurrency is introduced in these servers by simply spewing a large number of threads, each is identical and carries SIP requests all the way through the processing. We obtained much better understanding of the various processing modules, which we discussed in detail in this chapter.
With our detailed comparative study of M ∕ M ∕ 1 model with the experimental results with the open source SIP proxy server software (IPTEL 2002) and also with our proposed analytical M ∕ D ∕ 1 model, we continued our study more on the internal details of SPS software components, processes and complete software implementations. During the course of this research, we also investigated the limitations of the open source SPS software and replaced that with CISCO SPS software, which is designed, developed and implemented with latest software and hardware technologies for conducting our new set of experiments in the LAN environment. This chapter details the second part of our research in providing some reasonable solutions to the problems identified with our previous study discussed in Chap.4
The core idea for this part of research is to investigate the multi proxy environment, so that the SIP calls can pass through multiple proxy servers and latency is considered. We investigated the impact of Call Hold Time (CHT) in SPS performance with the help of a separate empirical study in the lab. For this study, we considered a multi-hop proxy server setup by redesigning our previous test setup with the latest IETF recommended trapezoidal SIP network architecture. The result of that investigation motivated us to further explore the SPS performance in LAN and WAN environment. The chapter provides details about our empirical study on CHT, results, empirical and predicted results on M ∕ M ∕ c based SPS model discussed in Chap. 6
Security to SIP network is of foremost important, since SIP is deployed in a large scale in various industries, universities, Department of Defense, banks, Wall Street and other areas where they use Internet telephony. Very little research has been focused so far in evaluating the SPS when different non-secure and secure transport protocols are used to transport the packet, while establishing a SIP session. This chapter is mainly focused on SPS performance overhead to the service providers when SIP calls are made using secured and non-secured transport protocols in the LAN environment.
In applications, it is often important to forecast the value of a random variable Y (which can be difficult or impossible to measure directly) using the observations of another random variable X (which can be more accessible for direct measurements). An important result of the probability theory is that when the two random variables are normal the forecast of Y in terms of X that minimizes the mean-square error is linear: $$\displaystyle{ Y ^{\prime} = a + bX }$$ (9.1)
This book focused on studying the performance characteristics and scalability aspects of SIP proxy server, which plays a vital role in the SIP architecture based network. Since it is not realistically possible to simulate or emulate the real world SIP network and study the performance characteristics of the SIP proxy server, significant effort was devoted in exploring new creative ways of designing the lab network and hardware setup. We conducted experiments with state of the art equipment, SIP proxy server software and other tools at the high tech CISCO “LASER” lab. We collected several data sets over the course of this research, did the comparative analysis and presented the findings in this thesis. The experimental results presented can be used as a benchmark data for the current and future SIP proxy server implementations. The performance parameters such as average response time or Call setup times, mean number of calls in the system, CPU utilization, Memory utilization, Queue size while establishing the SIP based media sessions is a very fundamental research in IP based telephony. Providing security for the SIP based network is a performance overhead for the service providers. We studied the performance impacts of processing SIP based calls are made using secured and non-secured transport protocols and presented the results.
Statistical definitions and formulae references from Matthews (2002) and Yan and Su (2009). ANOVA stands for Analysis Of Variance. In general, this statistical procedure compares the variance of scores in each of the groups of an experiment. With the two sample t-tests, we can have experiments with only two groups. With ANOVA, we can have as many groups as we would like to have.
... The IP architecture supports the carriage of any traffic (voice, video, and text) using best effort carriage devices. As reasonably priced as the IP network is, it poses some threat and challenges for Voice communication [2]. Tunnelling protocol such as Generic Routing Encapsulation Protocol provides packet-level encapsulation only with support for IPv4 and IPv6. ...
Article
Full-text available
Scientists succeeded in implementing conventional public switch telephone network (PSTN) into internet protocol by launching H.323 IP telephony. The irrelevant and unknown captions in H.323, computer scientists have replaced H.323 by Session Initiation Protocol (SIP) for Voice-over-IP (VoIP). However, the security of voice communication over IP is still a major concern. Besides, security and performance contradict features. VoIP exhibits a quality-of-service requirements that are sensitive to time. Example of such QoS requirements are delay, jitter, and packet loss. Integrating Internet Protocol Security (IPsec) with Generic Routing Encapsulation (GRE) encrypts and authenticate packets from the sender to receiver, but that raises the question of performance as VoIP is time sensitive. Consequently, three codecs were evaluated to determine the efficiency of each on GRE and IPsec implementation on Internet Protocol version 4 and Internet Protocol version 6 (IPv4 and IPv6), respectively. The topology design and device configuration in this study adopted Graphic Network Simulator 3 (GNS3) and Distributed Internet Traffic Generator (D-ITG) to generate VoIP traffic. The evaluation revealed that the G.723.1 codec achieved better results on IPv4 and IPv6 over GRE with IPsec than other codecs used in the experiment. Furthermore, the codec of choice is a major factor in IPsec VoIP deployment, as also revealed in this study.
... SIP is a signaling protocol defined by the Internet Engineering Task Force (IETF) for the establishment, release and modification of multimedia sessions [3]. SIP is used in the IMS as signaling protocol for session control and service control. ...
Article
Full-text available
IP Multimedia Subsystem “IMS” is the modular architecture that provides users with reliable access to multimedia services (voice, video conferencing, interactive services, etc.). However, quality of service and security are one of the main concerns of operators when it comes to multimedia services. The Multi-Protocol Label Switching (MPLS) protocol ensures these two basic needs. IMS is primarily based on the Session Initiation Protocol (SIP) for (i) routing, (ii) establishing, (iii) modifying, and (iv) closing a multimedia session. The SIP protocol represents the most influential protocol in the IMS architecture. The objective of this paper is to evaluate the performance of the SIP protocols in an IMS architecture with the MPLS protocol as the transport layer protocol. The evaluation will focus on: (I) the impact of the number of connections on the duration of SIP session establishment and (Ii) the impact of the number of users on the performances of IMS services (VOIP and Videoconferencing).
Article
A two-stage cultivation strategy was applied to mixed microalgae, which were first cultured in complete nutrient medium then switched to different nutrient-free mediums in order to assess the impact of nutrient starvation on intracellular biochemical components of mixed microalgae. The effects of nitrogen, sulfur and phosphorus starvation on cell counts, chlorophyll, carotenoid, protein, starch and lipid content of the mixed microalgae are compared in this study. The obtained results revealed that starch, as a dominant storage compound, was the highest in the nitrogen-free medium up to 49% of dry weight (DW). Protein and chlorophyll content declined slightly from 512 to 472.96 mg.l⁻¹ and 29.43 to 26.58 mg.l⁻¹ only in the case of nitrogen starvation. Nitrogen starvation showed the best result as it ceased cell division immediately whereby in the case of sulfur and phosphorus starvation, cell division was not interrupted as microalgae are a pool of phosphorus and store sulfur. Calculation of starch and lipid energy content on the basis of electron equivalent and J/mg dry weight showed the higher energy content of lipid in compare with that of starch for all nutrients starvation. Nitrogen deprivation condition represents the superlative energy value of storage compounds content with the value of 17.61 J/mg dry weight. The finding proves the potential of attractive and economically feasible mixed microalgae cultivation for high percentages of storage compounds production under nitrogen starvation
Article
Full-text available
Within integrated services digital networks (ISDN), all inter-change signaling messages for ISDN and intelligent network (IN) call controls are carried through a common channel signaling network (CCSN) as a backbone signaling network. Since CCSN usually have very strict reliability requirements as well as good performance objectives, performance and reliability of CCSN need to be jointly analyzed. This paper evaluates the mean end-to-end delay of a single-mated pair (SMP) CCSN for various call-arrival rates in a normal state and several failure states, as a performance index. As a performability index, this paper also analyzes the mean time to unreliable operation (MTUO) of a given network for various call-arrival rates at each signaling end point (SEP) and the failure rate of each signaling transfer point (STP). These results can be used in the design of common-channel signaling networks. This performability evaluation can be further studied for varying the failure rates of SEP and linksets, and the various threshold values of the unreliable operation
Conference Paper
Wireless networks beyond 2G aim at supporting real-time applications such as VoIP. Signalling protocols such as the session initiation protocol (SIP) are used to set up VoIP calls. The session setup time can be affected by the quality of the wireless link, measured in terms of frame error rate (FER), which can result in retransmission of lost packets and can lengthen the session setup time. Therefore, such protocols should have a session setup time optimized against loss. One way to do so is by choosing the appropriate retransmission timer and the underlying protocols. In this paper, we focus on the SIP protocol and evaluate its session setup delay with various underlying protocols (transport control protocol (TCP), user datagram protocol (UDP), radio link protocol (RLP)) as a function of the FER. For a 9.6 Kbps channel, the SIP session setup time can be up to 84.5 s with UDP and 105 s with TCP when the FER is up to 10%. The use of RLP(1,2,3) puts the session setup time down to 2.4 s under UDP and 3.5 s under TCP for the same FER and the same channel bandwidth.
Article
The next generation of enterprise networks is undergoing major changes as a plethora of new architectures, applications, and services begin to roll out within businesses. In general, the world of voice/telephony, video, and data are "converging" into a global communications network. The purpose of this paper is twofold. First, the design, analysis, and performance of a session initiation protocol (SIP)-based videoconferencing desktop client, which has been developed and deployed over Internet2, is presented. Second, a guideline for managing SIP-based services to be deployed within enterprises, which addresses several challenges in each layer, such as network address translator (NAT)/FW issues, directory service integration issues, and interoperability issues, is proposed. Several detailed experimental results related to interoperability and conformance that were carried out are presented. Findings of extensive SIP/NAT traversal analysis through network traffic measurements are reported. The lessons learned from both the design of a new SIP-based voice/video client, as well as management challenges with enterprise deployment are highlighted.
Article
The authors wish to provide a unification of this issue's papers by providing a clear context for them. Common channel signaling can be defined as the system that enables stored program control exchanges, network databases, and other nodes of a network to exchange messages related to call setup, supervision, and take down (call and connection control information); information needed for distributed application processing (interprocess query and response or user-to-user data); and network management information
Article
In a public switched telephone network (PSTN), the control information is exchanged among the network elements by signaling. The early telephone system used circuit associated signaling or inband signaling where the same transmission channel was shared by both signaling and user information (voice or data). Another signaling method is common channel signaling (CCS). Unlike the circuit associated signaling, CCS uses a separate out-of-band signaling network to carry signaling messages. CCS consists of supervisory functions, addressing, and providing call information. A CCS channel conveys messages to initiate and terminate calls, check on the status of some part of the network, and control the amount of traffic allowed. Signaling System No. 7 (SS7) is a CCS system developed to satisfy the telephone operating companies' requirements for an improvement to the earlier inband signaling systems. The advantages of CCS are listed