The transition from legacy transport networks using SONET/SDH to packet optical networks has not been an easy one. Access and aggregation transport facilities continue to be primarily SONET/SDH based even though IP/Ethernet payloads dominate traffic requirements. Some insights as to what the timing of the transition will be can be gained by analyzing individual customer segments.
Video services place the greatest traffic load on the network. The residential segment is the heaviest user of video and the one that shapes service offerings and overall customer expectations of service quality. Subscription based video services require at least 20 Mbps data rates to each household. The bandwidth requirement is driven by the widespread acceptance of HDTV and the need to support multiple TVs per household. High speed Internet service increasingly includes video content. Today broadband users typically expect at least five Mbps downstream and one Mbps upstream data rates in order to download streaming video and to upload videos to social networking websites. Legacy residential networks must be replaced entirely or substantially rebuilt to support video services. Carrier Ethernet facilities using IP DSLAM, GPON, or active Ethernet technology are reaching maturity as Triple Play service delivery vehicles as is IP/Ethernet for support of voice, video and Internet traffic.
In the U.S. cable TV operators enjoy a substantial lead over wireline telecom operators in delivering Triple Play services, however AT&T and Verizon have been reporting strong results in selling their Triple Play services, and the U.S. Stimulus program will help many rural telecom operators build out video services. Consequently, the residential segment will see substantial growth in Triple Play service usage. An important public policy issue remains, however, as what to do about the very large embedded base of copper loop plant and associated circuit switches that is still in place and how to continue to provide service to the significant population segment whose sole communications vehicle is analog voice.
The enterprise customer segment presents the biggest paradox. Ethernet and IP that are universally used for data services and VoIP based voice switches are widely employed for private voice switching. However, most enterprise sites are served by T1/E1 lines because they can be provisioned over ubiquitous copper loop plant. Even very large enterprises use many more T1 lines than high speed optical links due to the distributed nature of their workforces. The three most important applications of T1 access lines for enterprise have been for voice trunking to the PSTN, Frame Relay transport and Internet access. In general T1 lines are priced higher than IP/Ethernet alternatives (Several hundred dollars per month vs. $100/month for business class DSL.) Consequently, there is no big push by service providers to migrate enterprise T1 users to business class DSL. Even Next Gen service offerings are delivered using T1 access. For example, MPLS VPN-IETF RFC 4026 (formerly RFC 2547) is positioned as an upgrade for Frame Relay users and is the fastest growing enterprise service offering. However, most enterprise establishments use T1/E1 for MPLS VPN service because of the ubiquity of T1 and current bandwidth requirements.
Those enterprise sites that require high speed optical access services generally are limited to POS (Packet over SONET) transport services. Ethernet services such as VPLS are generally provided only at Edge Routers on a service provider's core IP MPLS network. Even the largest core networks have no more than thirty or so Edge Router nodes on their core networks. Consequently, enterprise traffic is backhauled for hundreds of miles using POS to reach the services edge of the core network.
Retail Carrier Ethernet services for enterprises are slowly emerging none-the-less. They are being rolled out initially in the financial centers of the very large cities that sit on the core network. A small number of retail Carrier Ethernet services produce enough traffic to overwhelm network bandwidth even when most end-users continue to use T1 (100 Mbps ports vs. 1.5 Mbps ports).
The final customer segment is wireless backhaul. It is the fastest growing customer segment. Once again Ethernet traffic will soon account for most of the traffic. As in the other segments the addition of 100 Mbps Ethernet used for 4G services will quickly overwhelm the T1 bandwidth currently used for 2G or 3G backhaul. Voice, however, continues to be the dominant mobile service revenue generator and operators are indicating that they intend to leave 2G networks in place to support it. Furthermore, 3G service is still being built-out in the U.S. and is a long way from the saturation point. Once again a pattern of very rapid growth in IP/Ethernet traffic combined with steadily increasing T1/E1 port requirements emerges.
This creates a messy problem for network architects in that they must support large numbers of T1 ports and SONET/SDH data streams even though the payload is overwhelmingly IP/Ethernet. I believe this is why Packet-Optical Transport (P-OT) solutions have created such a buzz among systems vendors. These solutions provide a hedge against service providers' inability to accurately time network transition. P-OT incorporates both time division and statistical multiplexing functions. This combined capability reduces the economic penalties of guessing wrong on the timing of network transition.
Michael Kennedy is a regular FierceTelecom columnist and is the co-founder and Managing Partner of Network Strategy Partners, LLC (NSP) www.nspllc.com -management consultants to the networking industry. He can be reached at [email protected].