 | Issue: | North America 2008 |
| Article no.: | 16 | |
| Topic: | Differentiated quality of IP service | |
| Author: | Jay Jayasimha | |
| Title: | CTO | |
| Organisation: | Veraz Networks | |
| PDF size: | 240KB |
About author
Mr Jayasimha has over 18 years experience in the areas of call processing, network planning and switching systems development. He has served as the Chief Technology Officer at Veraz Networks since 1999. Mr Jayasimha has been instrumental in the design and architecture of the ControlSwitch™, Veraz’ IMS-based switching product, and has made significant contributions in the areas of call processing, services and routing architectures. Prior to Veraz Networks, he held positions at British Telecommunication, MCI, Link USA Corp (now Global Crossing), and Qualcomm. Mr Jayasimha holds patents in the areas of call processing and call routing and is a member of IEEE/ACM. Mr Jayasimha holds a Bachelors Engineering (hons) degree in Electrical Engg/Computer Science, from Birla Institute of Technology and Sciences, Pilani, Rajasthan, India, and his professional credentials include IEEE/ACM.
Article abstract
The emergence of IP as a unifying technology across the telecommunications industry has fundamentally altered the way consumers and enterprises buy, use and experience telecommunications services. While this rapid, technology-led, advancement is likely only to accelerate in the future; today, service providers and their customers are grappling with fundamental business issues, the answers to which will likely be as important as technology in shaping the industry’s future.
Full Article
The last decade has seen explosive improvements in communication technologies, which have enabled the emergence of completely new service providers. These key disruptive technologies include voice over IP (VoIP) and all its derivatives, IT-oriented service delivery platforms, java based applications infrastructure, 2.5, 3 and 4G technologies, high speed Internet access via cable, satellite communications, and a plethora of multimedia handsets including the Apple iPhone. The combination of new non-traditional competitors and technology has blurred previously well-established lines demarcating communications providers, applications/service providers, content providers and media/entertainment providers. Often using novel business models (e.g. advertising supported), the new service providers are seriously challenging the traditional telcos one-size fits all, pay-per-minute approach. Within the network, there is an on-going movement towards convergence at all levels – devices, networks (access and core), services and soon data. There are several technologies taking shape that will enable data convergence that will be the next quantum jump in technology, thus completing the convergence picture. There is a fundamental common theme in this explosive growth that we have been and are continuing to see – the emergence of IP as the underlying network winner. One could see this coming in the early 90s when there was a shift to moving services from TDM infrastructures to an IP-based services infrastructure located in a call centre. The cost involved in bringing the caller to the service still remained, and the question was how to take the service to the caller. We could see IP networking emerge as the possible solution to this conundrum. The simplicity of IP is one of the major reasons for this large scale adoption. Obviously, with simplicity come several caveats. We have seen malware of several kinds hit networks. This required greater scrutiny into what caused these security issues, and networks are still battling intrusions with varying degrees of success depending on the type of the malware. Non-IP based access networks are moving rapidly towards an IP-based core network from which services are delivered. The trend has been towards the proliferation of IP-based access networks in the form of broadband over DSL, cable, more recently WiMAX and the upcoming LTE (Long Term Evolution). Though broadband networks are designed for data, voice still seems to be the application that service providers need to provide to make their business cases work. There have been several instances of Voice over DSL, cable and WiMAX deployments. Voice sessions are not as forgiving as data sessions of packet loss, round trip delay, jitter etc. and without care network designed primarily for data often provide an intolerable quality of experience for the voice services. On the other hand, even under optimal network conditions, if the number of voice sessions put through the pipe exceeds the bandwidth capacity, this could result in degraded quality of experience for all the subscribers over that pipe. Quality of service (QoS) then contributes significantly towards subscriber satisfaction when using voice over broadband services. Experienced VoIP providers have capabilities built into the core of their offering that use innovative algorithms with adaptive jitter buffers and resilience to packet loss to provide good perceived voice quality for the subscribers even in poor network conditions. There are two types of QoS models – the Absolute QoS model and the Relative QoS model. Relative QoS models are applicable in the core networks where abundant bandwidth is available. In this case, certain classes of traffic will have a better QoS than other traffic classes though there are no specific quantifiable guarantees. A well-known example is DiffServ where IP packets are classified and marked with certain codes using the TOS bits in the header. Routers then forward packets based on these codes which allows them to selectively minimize delay, maximize throughput, reliability etc. Absolute QoS models on the other hand provide more stringent controls over the bandwidth allocation, usage and policing of the resources per subscriber. This is applicable more on the access networks where there could be recognition of the service based on the subscriber’s session, and this session could have a prescribed traffic classification type with prescribed bandwidth allocation. Figure 1 shows an end-to-end QoS architecture with absolute QoS enforced at the edges and relative QoS at the Core. The figure is a very generic depiction where the access networks could be WiMAX (Access Service Network (ASN) gateways with the IP tunnels setup between the ASN and Core Service Network (CSN) for absolute QoS enforcement) or fixed line IP-Connectivity Access Network (IP-CAN) (DSL/Cable) etc. The Subscriber Session Resource database holds the bandwidth authorized for the subscriber and provides the details of the allocation of bandwidth per session type for the subscribers. The session types could be voice sessions, file share sessions, video sessions etc. Unknown session types will be classified ‘best effort‘ and provided the least priority. This type of advanced absolute QoS model can provide very granular control over the bandwidth allocation and policing at a session level, user level and at the network level providing dynamic control over bandwidth allocation and management. Figure 1 – Simplified representation of End to End QoS delivery Figure 2 – Session Level, User level and Network Level QoS Figure 2 shows the granular control over session level QoS. Each user session is admitted as long as the bandwidth allocated to the user is not exceeded. Dynamic changes in bandwidth to the user’s sessions may be applied. If the SLA of user 1 is higher than the SLA of user 2, then the priority of the user 1’s sessions take precedence over user 2’s sessions. The user level QoS is dynamically managed so that the network level QoS is preserved. This level of granular and dynamic control over the resource allocation, management and enforcing of QoS brings an enhanced service offering into the telecom service portfolio. This is significant value added by the transport pipe providers and provides a large revenue opportunity. It is not clear who will end up paying for these guaranteed SLAs, and how the business case will work out: will the telecom service providers with the walled garden approach provide this service to their customers? It remains to be seen as to who this paying customer will be: the end consumer, or the service providers or the content providers? There is a very complex set of forces at play and no objective logical conclusions can be drawn as yet as to what the winning business model will be. Eventually these complex forces will find their equilibrium and a verdict will emerge. Of course, the idea of paying for differentiated QoS goes against the sentiment of the free Internet and net neutrality proponents. However, with the explosive growth of bandwidth hungry applications, there may not be a choice over this issue. Today, bandwidth may be in abundance, but applications are getting more and more sophisticated demanding more and more bandwidth. Remember, it was not that long ago that a megabyte of RAM on the PCs was considered a lot. So, in a few years, bandwidth, which is considered abundant today, may in fact be a scarce resource. Telecom providers have provided reasonably good quality of service in the past and may provide a good quality of experience to the Internet users in the future. From the consumer perspective, the question of whether one would need a premium service to be able to casually use the Internet needs to be determined based on how Internet usage patterns evolve in the coming years. Gaming communities will definitely see the ability to get higher quality, differentiated service as an attractive proposition. The choice of premium service versus regular service should be offered to the Internet consumers, and it is up to the consumer to choose the option that works best for them. As a technology provider, it doesn’t seem to be right to take a stand on these issues; technology providers should rather be agnostic to these issues and provide the best products to enhance the consumers’ quality of experience.
