by Eve Griliches, ACG Research
(Editor's note: This article is also included in FierceTelecom's new eBook called Packet Optical Networking Platforms: Maintaining the legacy, next-gen balance. Click here to get your free copy today.)
Carriers are under intense cost pressures, and to maintain profitability they must introduce new services to replace declining revenues. Most of these new IP services are rich in media, sensitive to the quality of network issues, and extremely bandwidth hungry. As a consequence, providers are demanding fully integrated products that support optical transmission, TDM, and packet-based services.
This converged platform is now widely referred to as a Packet Optical Transport System (POTS) defined as including: a WDM portion with a reconfigurable add-drop module (ROADM); a TDM interface, which can multiplex and groom TDM traffic (preferably an MSPP on a blade); and an Ethernet switching interface, all supported in the same chassis. POTS products should support end-to-end differentiated services with statistical multiplexing and high quality traditional services, running on top of a flexible optical mesh. They often support some form of connection-oriented Ethernet and are required to support OTN and OAM capabilities. Now that we have defined the product type, let's discuss the different types of products coming to market.
Routers with optics
This approach begins with a highly scalable core router and integrates transponder optics thus decreasing the expense incurred when running between separate router and optical transport devices by removing a set of transponders at each termination site.
OTN switching must be added to the routing switch fabric, either as a separate fabric, or preferably within the same packet switched fabric creating a 'hybrid' which supports both container (OTN) and packet-based switching without latency tradeoff. While bringing optics into a router seems like a cost effective no-brainer approach, it does burden the router with additional expense, and long distance transmission in these types of products has yet to be adequately deployed. Also, many operators prefer separate control planes for their router and optical products, not a combination. Router vendors obviously favor this approach.
Packet integration approach
This is the more conventionally accepted approach, where OTN and packet switching is added to the optical platform. To date, OTN has been available assisting in the migration of the SONET/SDH traffic which still remains the bulk of the traffic in transport networks. However, adding packet switching at Layer 2.5+ has been challenging and optical vendors are slowly coming up to speed, but have been slow to deliver. Edge and core routers will likely remain as separate products in these core networks, and only some of the transit traffic (up to 20 percent) might be offloaded to the lower layers for switching rather than routing. ACG does not expect any reduction of core router revenue to be seriously affected by these products for at least 4-6 years.
The low down
Fully integrated products bear the brunt of hardship in several ways. First, it is inherently difficult to scale all the elements of the integrated product to a level which might be achieved with separate 'best of breed' products. Often one of the technologies simply does not achieve the feature parity or scalability goals initially defined at the outset. While reaching full feature and scalability parity is not impossible, it is simply a fact that some major Tier 1 providers have already pointed out as to why they are not pursuing the POTS integrated products quite as quickly as others.
Second, the cost goes up significantly as integration becomes more sophisticated and the software to control the product becomes more complex. And, the product(s) still need to be managed within and across the entire provider network.
Verizon is a perfect example of integrated POTS deployment. Verizon has deployed the Tellabs 7100 and Fujitsu 9500 in their metropolitan networks nationwide. Verizon now plans an entire long haul overlay with a larger POTS platform for their core transport network, expanding the use of the integrated product to their core network. Smaller tiered providers are also engaging with POTS products from vendors such as Cyan Optics and ECI.
But when we discussed this option with AT&T, the answer was clearly a scalability issue and limitation. To date, AT&T has not seen enough scale out of their optical platforms nor their routing platforms to commit to an integrated platform. Other European vendors are interested, but clearly want to move at a slower pace into the converged products domain and keep the lower layers together and slowly integrate packet functionality into their network.
There is another, albeit not an 'integrated' approach to the market, which is the simple 'best-of-breed' multiple product deployment approach. While this was the original and general delivery model, vendors who have stayed the course and kept products separate maintaining that best-of-breed is the way to go, have found that the packaging of their multiple products has resulted in significant discounting, hitting their bottom line and causing them to lose revenue even though they are shipping an increasing volume of product.
ACG believes this standalone product strategy is quickly falling from favor and will ultimately be eclipsed by integrated products such as POTS.
In the past 10 years requested "God boxes" were extremely unrealistic in scope, with few vendors ever really trying to approach the solution seriously. However, the POTS approach is a very possible and doable combination of technologies, and not at all an unrealistic request.
However, to deliver high quality and fully functioning products will not be easy, and each vendor will take different approaches, some with success, and some without success. Creating a POTS product requires architectural agility and expertise in transport design. It also requires understanding of the benefits TDM has had in networks to date, as well as the flexibility and increasing opportunity packet switching brings to the products. Not all vendors excel in all these areas, those that work hard to 'beef up' their engineering and technical management staff will be better positioned as this market really begins to take off in the next couple of years.
Eve Griliches is a managing partner at ACG Research. She can be reached at [email protected].