There is widespread consensus that access and aggregation networks will employ packet-based transport featuring 10 Gigabit Ethernet (10 GE) and higher speed interfaces. The consensus stems from packet-based transport's cost, bandwidth efficiency, and manageability advantages over the current generation of equipment featuring packet over SONET/SDH (POS) and DWDM technologies.
However, in the U.S. and elsewhere POS transport accounts for most facilities and there are a lot of obstacles to migration to a packet-based (Ethernet) transport vehicle. Access network reliance on copper-twisted pair continues to be a barrier for both residential and enterprise networks. Verizon with the largest U.S. Fiber to the Premises (FTTP) initiative reports 3.1 million FiOS subscribers versus 19.7 million residential switched access line customers. The enterprise picture is similar where business DSL, switched access lines, and T1 lines dominate access and nearly all of these run over copper. Even packet-based services to large enterprises such as MPLS VPN are usually transported over copper-based T1.
Most high speed enterprise services such as optical Private Line services are either directly offered as SONET or carried over SONET/DWDM infrastructure. This pervasive use of copper cable and SONET creates financial and operational barriers for both service providers and their subscribers. From the service provider perspective most of the cabling infrastructure is quite mature as is the Operations Support Systems (OSS) used to manage and maintain the infrastructure. Therefore, recurring costs are low and these services represent a major source of cash flow. Also infrastructure refresh must compete for funds with attractive alternatives such as wireless broadband and value-added enterprise services. Subscribers resist change because it disrupts their operations and they are wary of compromising availability and network security.
The expected slow migration from legacy services to Next Gen services such as Carrier Ethernet and Triple Play also poses challenges for network architects because legacy technologies must be supported for many more years. This is an especially big problem in the access network where large quantities of T1 will remain for a long time. For example, U.S wireless operators have indicated their intensions to stick with 2G for voice which uses TDM and circuit switched infrastructure even as they plan to rapidly move data services to 4G. Carrier Ethernet solutions that emulate TDM services encounter difficulties analogous to those that SONET solutions have in transporting packets. Migration to packet-based transport also must overcome service provider concerns about availability and reliability. Service providers see critical gaps in the areas of software defects, procedural errors, modular software patching and rollback, element management systems and fault management and root cause analysis according to a recent study by Network Strategy Partners.
Migration to packet-based transport will be driven by the take up rate for new bandwidth intensive services--especially video on the Internet, Triple Play video including Video on Demand and HDTV, movement of MPLS VPN beyond T1 speeds, and 4G mobile broadband. Internet video streaming is the most important driver of packet traffic due to the widespread adoption of the Internet by consumers and enterprises and due to better broadband services that offer multiple megabit per second download data rates as contrasted with sub megabit download data rates of first generation broadband.
Mobile broadband is another important driver. It is expected to grow by as much as 35 times its current volume over the next five years. This growth, admittedly off a very small base, is due to consumer friendly billing models combined with new devices and new technologies that are creating imaginative applications and increased usage. To the extent that these bandwidth intensive applications succeed the remaining legacy technologies and services will account for a minority of traffic and packet-based transport will become more cost efficient than the existing SONET/DWDM infrastructure. Obviously the supply-side also plays an important part of this analysis. Another way of predicting the rate of migration is to tie it to the rollout of packet-based access networks. Viewed in this way migration depends on the rate of deployment of systems such as Active Ethernet, GPON, IP DSLAM and optical Ethernet.
I expect a steady evolution toward Packet Optical Transport rather than a one or two year shift. Key gating factors are the dual take-up rate for bandwidth intensive services/packet-based access networks and systems vendors' progress in improving the reliability and end-to-end management of Next Gen equipment. A further factor that should aid the transition is the growing competence of systems vendors to act as both systems integrators and operators of turnkey solutions.
Michael Kennedy is 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].