Best of all worlds: intelligent hybrid platforms offer reliability of legacy, flexibility of next-gen systems

Telecommunications Americas,  May 15, 2002  by Michael Schneider

• E-mail
• Print
• Link

The latest buzz is that the faint pulse of telecom is beating in the metro networking market. Telecom managers know that long-haul fiber capacity is abundant and high-speed, last-mile access solutions are already being deployed at breakneck pace. So the sweet spot for revenue requires metro networks that pull it all together. It is the next piece needed to enable the wonderful gigabit services everyone's talking about. However, planners and engineers can no longer simply pull out the corporate checkbook and start building.

A lot of equipment is already in place. Maybe today's low price of legacy product is enticing enough to make use of this technology. But is it ready to handle the growing demand for advanced data-centric services like native Ethernet and Gigabit Ethernet for VIANs and VPNs? Can that architecture match up with the requirements of on-demand, wavelength services?

MANs are built typically with SONET rings made up of ADMs (add/drop multiplexers). Synchronous transport brought digital telephony into the optical era with more cost-effective and reliable network elements and topologies. DCS (digital cross-connects) are used to interconnect traffic from ring to ring. They perform a "groom and fill" function that helps make better use of backhaul circuits destined for central switching locadons or POPs. Recently, DWDM has been used in some metro applications to increase fiber capacity in much the same way as it is for long haul. It may be used to offer large-capacity, carrier-to-carrier connections in the metro or for fiber-exhaust in regional networks such as the New York/Washington, D.C., corridor. Some inexpensive versions of DWDM such as low-count lambdas over shorter distances, have been used for SANs (storage area networks) and other high-bandwidth, point-to-point applications. Combine these solutions and one could have a reliable, albeit disjunctive, multiservice met ro network. Unfortunately such a network is not scalable and flexible enough for new services models.

Trying to tap into one of the few growth areas, vendors increasingly are announcing "next generation" or "metro" products that promise to help carriers do it smarter, faster and better. However, many of these god boxes have failed to find favor with carriers. It is not optimal for next-gen transport elements to possess all the functionality of service layer voice and data switches.

Among these myriad options are also a few solutions that efficiently integrate the essential transport features for these new services into a very small, dense, flexible "hybrid-optical" platform. Topped off with more intelligence, these make up a cost-effective, scalable metro infrastructure geared to handle today's traffic as well as future data-centric services.

Figure 1 shows a typical metro network built with traditional network elements. There are stacked rings of OC-3, 12 and 48 with DCS for interconnection and grooming. Two types of WDM systems are shown: A regional system is connecting two POPs in different cities and an enterprise system is used for the SAN. More detail at the metro access shows the need for customer premises or access node equipment such as frame relay access devices, ATM concentrators and WAN access routers for service-layer aggregation.

Figure 2 shows a hybrid-optical platform that combines a multi-ADM function to close multiple rings and enable mesh topologies, with built-in DCS for grooming and integrated DWDM for both long-reach regional connections and short reach for SAN or OADM applications. Some platforms even integrate data handling such as GigE, FICON, Fibre Channel and video.

At the first stages of planning and implementation, both approaches may seem about equal in cost and complexity. Depending on how small the network is, the old way can even be less expensive. A simple eight-node, OC-48 transport ring for voice and some data services is half the initial capital cost of one built using the new hybrid-optical systems. Soon after the plans go from paper to plant, though, some big differences arise. At a hub or POP site, the combination of a few gateway ADMs, a DCS and possible DWDM nodes add up to 10 times the space, seven times the power and cooling and three times the cabling of a hybrid solution. For a colocation site, that amounts to about $18,000 per month for one site.

Long-term maintenance costs differ, too. Each network element brings additional points of failure. Common spare parts are unlikely across different platforms and different vendors. For one DS3 office jumper, three or four different circuit packs may be doing the job of one in a hybrid system. Newer, high-density cards that make up hybrid equipment share universal subsystems, so an entire site spares kit may consist of only 10 components. Network engineers trained on traditional equipment are easier to find, but more than one person may have to cover the full spread of equipment. All of this amounts to a more than 65 percent savings on average. For one small, six-node MAN that's nearly $8 million saved over a five-year period, according to Probe Research.