1 Whitepaper Getting the Most Out of Your Migration to SIP Trunking
2 Introduction As part of the migration to an all IP infrastructure, more and more enterprises are adopting Session Initiation Protocol (SIP) trunking, using it to replace expensive legacy and Integrated Services Digital Network (ISDN) trunk lines for voice service. SIP has become the Voice-over IP (VoIP) protocol of choice for signaling, providing significant improvements over legacy IP protocols and Time Division Multiplexing (TDM). SIP trunking offers enterprises easier infrastructure integration and management, as well as increased scalability. In addition, enterprises receive significant cost savings on infrastructure as well as local and long distance charges, rapid development and access to new services that leverage the flexibility of SIP, simplicity of running multiple services (voice, video, and data) over the same trunks, and better integration with the enterprises other business applications. With these benefits in mind, it is no wonder that SIP adoption is becoming increasingly prevalent. A recent poll of nearly 400 participants revealed that more than 62% had already deployed SIP trunks or were planning on deploying SIP trunks during the next year 1. In addition, according to separate research, 50% of SMBs are using or evaluating SIP trunking 2. However, while there are significant benefits to SIP trunking, adopting this relatively new technology is not without challenges. Both service providers and enterprises need to make sure that any issues are resolved before they affect their customers quality of experience. Therefore, it is crucial that organizations utilize comprehensive testing and monitoring solutions to assure that any problems are discovered and handled quickly and effectively. This paper outlines the benefits and challenges involved in migrating to SIP trunking and provides guidance on how best to maximize the return on investment in a SIP trunking environment. Why Migrate to SIP Trunking? Traditionally, telephony services were offered via a physical trunk consisting of multiple channels. With this model, adding more channels was costly and time-consuming. In addition, with traditional telephony services, it was complicated and sometimes even impossible to incorporate many of the newly available Internet-related services that provide a competitive edge. SIP trunking replaces physical connections with virtual ones, establishing sessions in an IP network that enable anything from a phone call to a multi-media conference. Bandwidth can easily be increased without necessitating a physical overhaul and SIP can be utilized for a wide variety of existing services and applications, such as , web browsing, and fax. As a result, enterprises can easily tie businesscritical applications together, simplify access for remote employees and agents, and leverage a single network connection for services such as voice-enriched e-commerce, Instant Messaging (IM), conferencing, application sharing, video, data sharing, and more. In today s economy, budgetary constrictions are also a powerful influence in inducing enterprises to adopt new technologies. Thus, it should come as no surprise that the promise of lower operational and capital expenses is driving the adoption of SIP trunking in the marketplace 3. The virtual aspects of SIP trunking eliminate the need for businesses to purchase costly connection equipment such as gateways connecting corporate environments to Public Switched Telephone Networks (PSTNs). It also lowers deployment and call costs (local and long distance), as well as provides an easy and cost-effective way to add new unified communications and other services down the line. The technological benefits combined with the potential significant costsavings makes it very compelling for enterprises to implement SIP. In fact, using SIP trunking in place of older trunking technology such as Primary Rate Interface (PRI) can produce savings of 50% or more 4. Facing the Challenges As with most technologies and perhaps even more so with SIP, since it is a still maturing technology there are many issues that must be navigated in order to realize all of the benefits mentioned previously. These challenges include evolving standards, assuring Quality of Service (QoS), integrating old and new infrastructure, scaling appropriately, and creating a
3 secure environment. Lack of Dependable Standards SIP has been designed for maximum flexibility, making it an ideal platform to integrate numerous technologies. At the same time, these products are not all necessarily able to interoperate. There are multiple Requests for Comments (RFCs), draft specifications, and optional elements available for SIP and there are often places where standard definitions are open to interpretation. Furthermore, some organizations develop their own variants of SIP, such that strict adherence to RFCs will not guarantee interoperability. As a result, at Empirix we recommend interoperability testing instead of RFC-based conformance testing. Interoperability issues can result in calls not connecting properly or poor voice quality scores that are often very difficult to diagnose. It is impossible to guarantee successful equipment integration and service turn-up without testing. To help address the problem of incompatible SIP implementations, the SIP Forum has launched SIPconnect 5, an industry-wide, standards-based approach to ensuring SIP trunking interoperability. Focused on sticky RFC issues and overlapping recommendations, SIPconnect provides a common reference point for SIP trunking components, especially IP Private Branch Exchanges (PBXs), as well as services that equipment vendors, service providers, and enterprises can utilize. Even with this continuing effort, however, interoperability is an ongoing issue. Quality of Service Implementing the right QoS settings in conjunction with using the appropriate transport protocol is critical to maintaining superior voice quality. Transport protocols can be tricky in SIP trunking implementations with clear reasons to implement User Datagram Protocol (UDP) and Transmission Control Protocol (TCP). The prime benefit of UDP is transport speed. TCP is bandwidth-intensive and not fault tolerant, but provides guaranteed delivery and is required for security implementations when using SIP over Transport Layer Security (TLS). One example illustrating the importance of the above is a testing engagement Empirix conducted for a customer deploying IP in its enterprise. The Enterprise customer copied its existing data service policies and used them for its new voice services. Once the customer turned up its SIP trunk, which was running over UDP, the organization experienced poor voice quality on a number of lines. Testing revealed the issue to be that UDP traffic was being throttled based on the organization s data QoS settings, which were programmed to reduce bandwidth utilization of non-work related media (e.g., Youtube). As this example indicates, voice and other real-time applications often require their own, distinct QoS settings. IP and packet switched networks were not designed for real-time communications. The introduction of latency and jitter that might be acceptable for data applications can be destructive to voice, video, and other real-time applications. This issue can be further exacerbated in a SIP trunking environment where other data services are using the same pipe into the enterprise. Infrastructure Integration Enterprises often slowly transition to IP and SIP. This situation produces a complex environment in which TDM infrastructure and endpoints as well as legacy IP voice protocols still need to be supported and integrated, even when deploying SIP trunking services. In other words, traditional voice, enhanced voices services such as voic and Class 5 features, as well as FAX, video, messaging, and other value-added services all need to perform flawlessly within the same environment. While it may be true that specific technologies work fine individually, there is no guarantee that they will continue to work within an integrated environment. It is therefore crucial that thorough testing is performed and all interoperability issues are resolved before an integrated environment goes live. Scaling Of course all organizations migrating to SIP and especially service providers want a fast and efficient turn-up and accelerated time-torevenue (TTR). However, it can be difficult for service providers to access portions of the customer s network for verification. Thus, it can be challenging to ensure that the SIP trunk works all the way to the demarcation line between the service provider s network and the enterprise customer s environment.
4 A second obstacle that affects both service providers and enterprises is the allocation of bandwidth for the desired dial plan. If the environment is not set up properly, it could lead to a degradation of voice quality as simultaneous call volume scales. Security Ensuring security and preventing unwanted access remains a critical component of rolling out a SIP trunking service. SIP trunks may be delivered as public services coming into the enterprise, yielding the same issues as IP data connections: hacking, eavesdropping, and connection hijacking. These security issues are primarily addressed by the utilization of Session Border Controllers (SBCs), as well implementing Intrusion Detection/ Prevention Systems (IDS/IPS) and Deep Packet Inspection (DPI) devices. Such devices must be thoroughly tested before being utilized in a live setting in order to assure that they work in the specific enterprise environment, providing the required levels of inspection, detection, and prevention. Meeting SIP Trunking Migration Objectives Enterprises often establish Service Level Agreements (SLAs) requiring that certain objectives are met. While these types of SLAs are still in their infancy, they are becoming more widely adopted as SIP trunking service deployments continue to grow. However, it is important to note that while some SIP-related offerings are more established, many of the services are still maturing, changing, and being extended and it is not clear that SLAs will change and mature at the same rate as the service provider features. If a failure occurs, an organization must be able to troubleshoot down to the root cause, isolate the problem, and determine problem ownership. Although verifying SIP trunks prior to turn up helps establish SLA levels that can be referenced when problems occur, access to the right points in the network for testing and ongoing monitoring in both the service provider and enterprise s network is absolutely required. To help ensure service quality, enterprises must consider utilizing testing tools that look at the entire environment, including applicationlevel performance. Businesses must also keep in mind that ongoing monitoring and management are crucial especially as new features are added in order to ensure that the enterprise continues to run smoothly. From a service provider perspective, the overarching goal when offering and rolling out a SIP trunking service is to create a new revenue-generating program that will provide higher quality voice service and more value in terms of access to new, robust services for enterprise customers. During a rollout, it is critical that service providers prove the benefits of making the transition to SIP trunking by establishing a higher level of quality than traditional TDM voice service offerings. Simultaneously, they must ensure the reliability and scalability of any new services, while also keeping the deployment cost-effective and efficient. This balancing act requires that service providers be certain that all of their offerings are able to integrate seamlessly. In addition, carriers need to easily and cost-effectively verify the provisioned capacity and establish an SLA baseline of SIP trunks. Automated testing tools can help verify baseline QoS metrics that can be utilized in SLAs. Steps to Take for Testing the SIP Trunking Environment The recommendations in the following section are based on the assumption that the service provider s network functions properly, its core components work, and the network delivers superior quality of experience for voice services and enhanced services via SIP trunks to other customers. In other words, the service has already been successfully employed and it utilizes a mature testing and monitoring methodology. For the enterprise, this section assumes that organizations reading this paper currently maintain high quality voice capabilities with existing TDM and/or IP infrastructures. With that in mind, here are the steps to take in order to assure that networks work together to deliver superior service quality over SIP trunks. Establish Realistic Baselines The first and most important testing requirement is the ability to emulate real users and realistic scenarios. It is critical to establish a baseline for voice quality for single calls, for a sample of calls with full background load,
5 and for the maximum number of simultaneous calls. These metrics can be used for establishing a baseline for an SLA and for comparison down the road for SLA adherence. Of course, users do not always call in singly, nor do they necessarily call in at a consistent rate, and many quality problems occur at capacity, not with just a single call. Therefore, one of the crucial aspects of creating a realistic testing environment is the ability to scale to meet maximum call expectancy. As that kind of testing can be time-consuming and costly and often impossible through manual testing, it is suggested that companies utilize an automated test solution that can scale to meet the provisioned capacity for the SIP trunk. The initial step is to configure the test setup so it emulates the enterprise, driving test calls into the SIP trunk and then into the service provider s network infrastructure. To accomplish this goal, first, generate a single call and then ramp up to maximum call volume from the emulated enterprise. Areas to test can include end-point registration, call connectivity, voice quality, Dual-tone Multi-frequency (DTMF), fax, and video. The test solution utilized should also offer the flexibility to emulate different call flows and different IP PBXs. As mentioned previously, SIP is not implemented the same way at each network equipment manufacturer (NEM), so IP PBXs can easily vary in terms of how they support different call features. Add and Test Components Step-by- Step In order to gain a true understanding of the way in which each component works within specific enterprise environments, each piece should be added and then tested separately. Begin by adding real enterprise network components one at a time, starting with the SBC and continuing with the IP PBX. This step-wise approach will help verify the service provider s network without any enterprise components involved and then enable users to methodically test each enterprise component all the way out to the desktop. Along the way, a variety of test cases should be utilized, including looping scenarios in which cases are repeated to verify stability. Longevity test cases also help assure stability. These tests are run for multiple hours, overnight, or during the weekend, in order to uncover latent or memory-related problems. If different SIP trunking CODECs are being utilized for efficiency reasons or device compatibility requirements, then each of these must also be tested separately. Lastly, any existing or new services that can be run in combination with the SIP voice services should be tested (for example, voice conferencing along IM or 911 services with significant background loads). There are a number of setups for SIP trunk testing that provide different levels of coverage and insight into failures. For instance, calls can be set up to run in loopback mode, where they are looped back on a switch within the service provider s network and terminated on another emulated endpoint back at the same enterprise facility. This method provides a central point for test execution, monitoring, and evaluating results. For this scenario, be certain that the test and network are configured so that the Real-time Transfer Protocol (RTP) actually leaves the enterprise facility, in order to provide a true measure of voice quality. Other test scenarios include running tests from one enterprise location to another in order to verify intraenterprise call quality or sending test calls from the enterprise to a terminating test endpoint in the service provider s network. This second example provides a single path measure of call quality and a way to tie test results into the service provider s network monitoring solution. Continue to Monitor Identifying issues before customers or agents find them is essential to ensuring a superior quality of experience and minimize customer churn. Of equal importance is the ability to track applications and systems to predict where issues may occur in the future based on usage patterns, response or queuing times, and IVR performance. As a result, monitoring, performance management, and accurate as well as meaningful reporting are vital to long term service and application quality and reliability. Passive and active solutions are recommended so as to continuously monitor call quality and reliability, as well as provide a means for problem isolation, analysis, and diagnosis.
6 In service provider networks, solutions are typically distributed so they can access call information from various points in the network and provide correlated Call Detail Records (CDRs) back to a main network management system for monitoring and reporting. Enterprise solutions are usually aimed at monitoring the performance of contact centers and unified communications applications. Methods for Measuring Voice Quality Voice quality is arguably the most important measurement when it comes to successful SIP trunking implementations. There are a variety of algorithms that can used to obtain voice quality scores. Mean Opinion Scoring (MOS) is a subjective voice quality scoring methodology, specified by the International Telecommunications Union (ITU) in The measurement is based on a panel of human listeners and the scoring ranges from one to five. Since the algorithm is based on subjective feedback, trying to replicate it for automated testing or monitoring is obviously impractical. As a result, equivalent type scores in automated solutions are often derived from passive R-factor or active Perceptual Evaluation of Speech Quality (PESQ) voice quality metrics. These scores use a formula that takes into account both user perceptions and the cumulative effect of equipment impairments to arrive at a numeric expression of voice quality 6 and can provide high-level quality ratings and breakdowns of any issues affecting quality. Passive voice quality testing utilizes the E-model algorithm (a computational model for use in transmission planning 7 ) and leverages single-ended packet-based statistics to produce an R-factor voice quality score on a scale from 0 to 100. This measurement is non-referenced-based and does not open up packets to look at the contents. Thus, it is best suited for IP-to-IP test scenarios where transcoding is not involved. PESQ follows ITU recommendation P.862 and is the standard objective scoring methodology for active voice quality testing. PESQ methodology involves sending out speech clips and receiving them at another location. PESQ utilizes real voice prompts to measure the impact of VoIP infrastructure and any associated network issues on voice quality. For example, a test prompt, which is a real speech sample, could be sent from one emulated test point to another. The PESQ algorithm would then compare the speech sample that was sent out initially to the one that was received after traversing the VoIP network, producing a score on a scale of -0.5 to This active testing solution can be used for any SIP trunking test scenarios as long as each of the emulated test endpoints uses the same test prompt for comparison. Troubleshooting As mentioned previously, once SIP trunking has gone live, it is vital to continue to monitor the environment and catch any issues early, before they affect end users or employees. Troubleshooting and diagnosis can be a complex task in an IP environment, let alone in today s hybrid IP and TDM environments. Implementing a methodical approach to troubleshooting is therefore critical to reducing variables, isolating faults, eliminating network problem ownership debates, and preventing end users from experiencing problems. The most feasible method for troubleshooting in an enterprise is to take a layered approach. Initial focus should be on systems and infrastructure, in order to gain an understanding of performance and availability. Manual techniques for discovering issues can be unwieldy and cost-intensive, but advanced monitoring solutions provide alarms or indicators to highlight instantly where and when issues occur. If the infrastructure is sound, the next step is to investigate the applications and back-end office functionality that are deployed on the network. These should also be constantly monitored for latency, call failures, transaction issues, and other application performance or availability problems. The focus should be on IVR, CTI, Voice Activity Detection (VAD), and other parts of the application. Finally, if an issue still has not been resolved, it is time to research whether or not it is affecting the customer and agent experience. Empirix Hammer Test Solution The Empirix Hammer Test Solution can emulate real-world scenarios, providing complete validation of application performance with load, regression, and interoperability testing. It also provides customer
7 experience assessment and media quality analysis to preempt businessimpacting issues and maximize return on investment. Automated test capabilities ensure that SIP trunks are ready for turnup and provide a rich set of metrics (including voice quality scoring to use as a baseline for SLAs with enterprise customers). As a result, organizations can provide superior service quality and avoid post turn-up surprises and problem ownership questions. The solution is easy to set up and highly scalable for both enterprise and carrier-grade deployments. Conclusion SIP trunking is receiving widespread attention and adoption because it provides so many benefits. At the same time, organizations adopting this technology must overcome numerous obstacles and challenges. Utilizing the best practices discussed in this paper enables organizations to mitigate these challenges and allows them to gain the true benefits and savings of SIP trunking. problems quickly and efficiently. Thus, organizations can be confident that their SIP trunking migration will be cost-effective and worthwhile, and that issues will be identified before they affect customers resulting in a high quality customer experience. 1 Acme Packet SIP Trunking Webinar, No Jitter, June 24, 2009, 2 SIP Trunk Adoption Ahead of SIP Security Adoption, Nemertes Research, Spring, 2009, adoption_ahead_sip_security_adoption 3 SIP Trunking: IP s Cheapest Delivery Package, Light Reading s VOIP Services Insider, November 7, 2008, 4 Acme Packet SIP Trunking Webinar, No Jitter, June 24, 2009, 5 SIPconnect Forum, sipconnect 6 Method and Apparatus for Measuring Voice Quality on a VoIP Network, by Prakash Khanduri, 7 Question 8/12 E-Model Extension Towards Wideband Transmission and Future Telecommunication and Application Scenarios, International Telecommunications Union, ITU-T/studygroups/com12/sg12-q8.html Testing and monitoring a SIP trunking installation can help both enterprises and service providers. Enterprises can quickly realize cost savings, access new services, and provide broader user access, among other things. In addition, they can establish more comprehensive SLAs and better assure voice service quality. Service providers can accelerate TTR and reduce rollout costs. Both enterprises and service providers can ensure superior SIP trunking service quality as well as diagnose and isolate For a complete list of offices worldwide, or to find an authorized distributor in your area, please visit: /contactus Empirix. All rights reserved. All descriptions, specifications and prices are intended for general information only and are subject to change without notice. Some mentioned features are optional. All names, products, services, trademarks are used for identification purposes only and are the property of their respective organizations. Empirix and the star symbol design are trademarks of Empirix, Inc., Billerica, MA CP:WP:GTMOOYMTST:1212