Putting VoIP to Work: Softswitch Network Design and Testing: Softswitch Network Design and Testing

Description

• VoIP and softswitching: technical challenges, proven solutions
• State-of-the-art testing techniques for ensuring superior QoS
• Key design approaches for MPLS and differentiated services
• Covers protocols, topologies, and call flows
• By the author of the best-selling IP Telephony

The in-depth, up-to-the-minute technical guide to developing and deploying IP-based telephony.

This is a complete, in-depth technical guide to the challenges associated with building and deploying Voice over IP (VoIP) networks using softswitch technologies—and today's best solutions. Bill Douskalis, author of the best-selling IP Telephony, reviews the entire current state of the art in protocols for signaling, media transport, and network engineering—and presents expert guidance on designing telephony solutions that meet the needs of both carriers and customers. Coverage includes:

• New scenarios for design of VoIP networks based on Class 4 and Class 5 softswitch network technology
• Designing for performance and interoperability: challenges, examples, and comparisons of potential solutions
• Protocols of the new converged networks and design requirements for seamless integration of softswitch applications with the PSTN
• Call flows for SIP, H.323, MGCP, and Megaco
• Testing tips to ensure consistently high QoS in voiceband applications
• MPLS and IP differentiated services in VoIP networks: key roles and integration issues

Nobody has more experience implementing VoIP in large-scale networks than Bill Douskalis—and no book offers more insight for real-world VoIP design, construction, and deployment.

"The book is very well written and provides the practical information on all aspects of a VoIP network in the ever-changing field of communication technologies."

—Dr. Tushar Bhattacharjee, AT&T, Polytechnic University of New York

Extras

Related Articles

Protocol Interoperability and VoIP Performance

Testing Challenges in Packet Telephony

Sample Content

Table of Contents

Foreword.


Introduction.


1. Protocols and Topologies of the New PSTN.

The Continuing Evolution of the VoIP Protocol Stack. H.323: More Efficient Ways to Signal, with a Common Denominator. MGCP and NCS. SIP (RFC2543). The SDP Protocol: RFC 2327 with Extensions in RFC 2848 (PINT). Codecs, Tones, Faxes. H.248: The Megaco Protocol. SCTP: Stream Control Transport Protocol (RFC2960). Transport Adapter Layer Interface (TALI). Signaling Transport User Adaptation Layer Protocols: M3UA, M2UA, SUA, and IUA. Firewalls. Other Protocols.

Wireline, Point-to-Point Access Network. Service Provider End Office. Concluding Observations. Signaling, Media, and Platforms: The Necessities.

The Basics.

From Best Effort Service to Predictable Performance. Concluding Comments.

Protocol Overview. Registration, Admission, Status (RAS). Call Setup and Termination. H.323 over UDP. Ad Hoc Conferencing. Feature Call Flows. Firewalls.

Protocol Overview. Media Gateway Initialization and Registration with the Softswitch. Call Setup and Termination. MGCP-ISUP Interworking. Special Call Flows. SDP Protocol Review.

Protocol Overview. Registration with the MGC and Initialization. Call Setup. Extensions and Enhancements to Basic Megaco Signaling. Concluding Remarks.

Protocol Overview. Basic Signaling with SIP Endpoints. SIP-PSTN Interworking. Compliance and Failure Scenarios.

PSTN Protocols over IP. SCTP. M2UA. M3UA. Transport Adapter Layer Interface (TALI). Fax Relay in IP Networks. General Observations on Testing Issues. Keeping Track of the Objective. Direction in LAN Equipment Infrastructure.

Preface

Preface

It has been a little over a year since we looked at the evolution of the new public telephone network, and things have moved quickly enough for us to take another look at our progress and current direction. Indeed, the move towards adoption of IP-based integrated services by the service providers has accelerated over the last several months, from the perspective of creating an IP-based infrastructure for consumer and business applications. At the same time, a lot has been learned, and is continuing to be learned from the technical issues and testing challenges we have had to overcome in the design of VoIP networks and services.

Some say there are signs the business drivers for the new technology may not be materializing as quickly as industry analysts had hoped and expected. The so-called "killer application" that will capture the fancy of the consumer and will generate wonderful revenue streams for the service providers is not staring us in the eyes. Maybe the reason is because the enabling technology itself has not been ubiquitously deployed yet. Instead we seem to be operating in a space of multiple business opportunities, which require careful assessment of provider capabilities versus customer needs to identify the few business opportunities that will turn a profit. The promise of integrated services technology remains as exciting as ever, but it is also becoming more and more complex as we learn in detail about what we really need to deploy in order to offer viable—both technically and economically—services to the consumer. We are in the process of moving from network architecture and feasibility studies to bringing real solutions to the wall socket in the customer premise, and any conclusions about the success of the new and converged public network are still premature.

We have learned over the past year that network security, privacy, and traffic engineering are of paramount importance and not as easy to implement as had been originally expected. We always had the feeling this was the case, but we have now experienced glaring examples of security breaches in well-managed corporate networks, security attacks in the form of denial of service, and perpetual hacking attempts, and we have struggled with the subject of offering quality of service to the consumer in a manner that can be guaranteed in a contract.

The theory behind each piece of technology of the new networks is good and getting better, but putting it all together in a working product is a daunting task. Fortunately, the standards organizations have continued to formalize protocol specifications, thus bringing some predictability and stability in the transition between network architecture and implementation. However, we have also seen a continuing stream of new draft proposals to address follow-on issues arising from the adoption of those brand new standards. Again, this has always been the case to some degree for every new technology, but the magnitude of the redesign of the PSTN highlights the level of effort that continues to go into settling the specifications upon which equipment and the network topologies that employ them are being built. Furthermore, there are major areas of the new technology that have not even been addressed to the point where services can be reliably deployed. Starting from simple telephone conferencing between parties using solely VoIP technology across heterogeneous autonomous system boundaries (i.e., different service providers), and ending with the all-promising video teleconferencing service and multimedia collaboration, there is a significant amount of work waiting to be tackled.

One thing is certain-the new packet-based telephone network does not have a clear winning protocol for setting up a simple telephone call. We have seen the resilience of H.323, with all its incarnations, implementations, complications, and alleged flaws, and the proliferation of SIP in call signaling and in support areas deep inside softswitch platforms. On the other hand, MGCP is now both a formal standard and a major protocol of choice between the softswitch and the Media Gateway/Integrated Access Devices (MG/IAD) on the customer premise. The Megaco protocol is very elegant and seems to be addressing some interesting problems, but its timing is late and its rate of adoption will depend on whether business drivers are strong enough to dislodge what has been designed into the deployed equipment already.

Looking towards the core of the PSTN, we are now attempting to replace the SS7 signaling network with IP-based methods used to transport upper layers of the SS7 protocol (i.e., SCCP/ISUP/TCAP and above) over packet infrastructures. For those who are skeptical about the security ramifications of such an undertaking, we can all rest assured this issue looms large in the minds of the service providers and, by extension, the equipment manufacturers. A lot of work is going into addressing security problems in every aspect of VoIP telephony. The good news is that we seem to have accepted the use of the upper layers of the SS7 protocol (e.g., ISUP, SCCP) and the focus is on the technical effort to integrate those upper layers with the packet infrastructure of the network layer. What does this mean in simple terms? As an example, if you did global title translation in your old PSTN network, you will most likely still be doing global title translations in the new model, using most of the same TCAP messages, but this time transported over an IP infrastructure. From the viewpoint of robustness and reliability, this is a much less risky approach than defining an entirely new public signaling protocol from scratch, and if you are familiar with SS7, you are almost halfway there. Finally, equipment using the GR-303 and V5.2 protocols is also breathing new life to the TDM access network.

This year has brought us important technical breakthroughs that have demonstrated the commercial viability of optical switching in replacing traditional packet routing methods and optical-to-electrical conversions in the core network. Among the promises of l (lambda) switching are much higher bandwidth and migration to mesh optical topologies that could revolutionize core network design. Of course, let's not rush to exuberance too quickly, because SONET rings are not quite obsolete yet, and a lot of money is still being spent in mainstream traditional optical network installations using SONET/SDH. But pressure will be mounting in the next few years to develop reliable backbones with designed-in scalability, high bandwidth, and invariant performance characteristics of the transported applications, while offering protection switching from link and equipment failures. One nice and obvious benefit of optical switching in the core is the ability to use some form of label switching from the edge of the network, where signals are still in electrical form, through the high-speed backbone and out the other end, by somehow associating labels to lambdas, thus maintaining the utmost end-to-end performance for critical customer applications. When we consider the impact on VPN construction using optical means, the list becomes long and very exciting, with many technical and business benefits all around.

We are all aware of the quest for "five nines" reliability. We must keep in mind that this is an end-to-end figure, and not simply the reliability of a box in the network, and it is hard to predict the network reliability by just looking at the specifications of a box. There is much more to achieving the "five nines" than designing a fault tolerant switch, gateway, or router. Reliability is inseparable from the subject of network testing methodology and predicting network behavior in the presence of failure scenarios. I recall spending many hours on these two topics during the specification phase of a large VoIP design. The first temptation to conquer is the tendency to assume reliability and failure accountability is something that can be inserted in the network at a later date and thus should not be used as a reason to delay deployment. This notion could not be further from the truth. The rule of thumb is, the more complex the technology, the more diligent we have to be in the accounting of failure scenarios. The word "accounting" here is meant literally-the process of enumeration of failure scenarios is an accounting process. Those failures that have been accounted for in the design phase will have a higher probability of graceful and predictable handling, but the rest of them will cause trouble.

In this text the approach is to place the current state of the art in signaling and media transport on top of a reference topology that would require a multitude of specifications to be correctly implemented in order to function properly. Many of the protocols will need to interwork simultaneously, even for the simplest transactions. In this context we will discuss the infrastructure simple calls, conferencing, requirements and interoperability, interfacing with the PSTN, replication of existing Intelligent Network features and services, new telephony services, as well as issues associated with the underlying infrastructure. The reader must be familiar with TCP/UDP/IPs, as we will also cover the progress that has been made in devising a more streamlined, connection-oriented, reliable transport protocol. This new protocol, Stream Control Transport Protocol, or SCTP, will be discussed in the last chapter.

It is impossible to cover the entire technological breadth of an IP-based integrated services network in a single text, but the hope is that this effort will shed some light on the major issues facing key areas of this still very new technology. The intent is to provide a good reference for architectural decisions made by designers of networks and services. I hope you enjoy the book!

Bill Douskalis
April 2001

Updates

Submit Errata