AT&T Co, Stratacom Inc and Cisco Systems Inc got together on Tuesday for the most important Asynchronous Transfer Mode joint announcement seen to date. Both AT&T Network Systems and Stratacom announced Asynchronous Transfer Mode switches – the former has the large GCNS-2000 switch, with a total throughput rated at 20Gbps. Stratacom unveiled its 9.6Gbps BPX switch. At the same time AT&T said that it would be talking about its first Asynchronous Transfer Mode offering, dubbed Interspan ATM services within the next few weeks. This will be built on a mixture of its own switches, acting as the backbone, and Stratacom devices around the periphery. As such, AT&T becomes the first customer for the Stratacom BPX. The whole lot was capped by the news that AT&T, Stratacom and Cisco Systems have set up a joint development pact, to work on Asynchronous Transfer Mode interoperability and ATM-Frame Relay interworking.
Strategy
Stratacom’s strategy for entering the embryonic market is particularly cunning, since it aims to build on the existing penetration of its IPX cell/Frame Relay switches. For its new switch, Stratacom has moved from a bus-based architecture to one based on cross-point silicon bought in from a company that it does not want to name. The result is a device with a claimed total throughput of 9.6Gbps, presented to the outside world via 12 800Mbps slots. The company says that it is building on the experience gained through its IPX switch which uses a proprietary cell-switching system with a 24 byte cell size to carry interswitch traffic. While many Stratacom customers stick a standard Frame Relay interface in the switch, most also use the native cell protocol to mix data and voice in a pseudo-Asynchronous Transfer Mode network. This proprietary, cell-based approach has been the subject of criticism in the past on the grounds that it makes IPX switches difficult to interwork with (a charge that the company denies). In this case, however it is proving a strength, with vice-president of marketing Chris James saying that the company has valuable experience in cell, technologies which its competitors lack. A variety of factors make the BPX one to watch. Perhaps most important is the the claim of seamless integration with the IPX. This makes it attractive for users and providers with IPX-based networks, who find they are running out of steam, or want a planned migration to Asynchronous Transfer Mode. In particular it may appeal to public network operators who can add backbone capacity to their Stratacom-based Frame Relay networks, secure that the same boxes can be used for an Asynchronous Transfer Mode service.
By Chris Rose
It is, after all, one reason that AT&T plumped for the product. To ensure the IPXs and the BPXs work together nicely, Stratacom has just given the IPX an Asynchronous Transfer Mode trunk interface. On the BPX side it has built in the same useful, proprietary additions, such as cell prioritising, congestion and bandwidth management already found in the IPX. The fact that they are proprietary means that as soon as the traffic is passed to another manufacturer’s switch through a standard ATM-switch to ATM-switch interface, called the Network-to-Network Interface, NNI, the benefits are lost. However one area on which AT&T and Stratacom are working is an enriched Network-to-Network Interface that will incorporate at least some of their proprietary features. Neither will say much about their work in this area, but AT&T Network System’s product line manager for high-speed data communication services Jayne Fitzgerald says that I would like to see the very kind of thing that we are doing here drive the NNI standards process. Since the standard Network-to-Network Interface is still at its formative stage, this sounds plausible assuming they don’t mind presenting potential competitors with their crown jewels. Stratacom’s other point of differentiation is the variety of lower speed interfaces that it is planning. In addition to straight Asynchronous Transfer Mode connections, the BPX will eventually boast Fr
ame Relay, PABX, and other standard data and video interfaces. Frame Relay is promised both at conventional speeds, for connection to routers, but also at 45Mbps T3/E3, for multiplexed services. The first shipment, due in the fourth quarter, will include the high-speed Asynchronous Transfer Mode trunk services, the narrowband interfaces, and interworking with the IPX box.
Mix and match
The actual Asynchronous Transfer Mode user interface is scheduled for the second quarter of next year. This will enable the company offer full ATM-Frame Relay integration, with users able to mix and match the two technologies between sites. How does the company manage to get the variety of slow-speed interfaces? Simple: a fully configured BPX is essentially an Asynchronous Transfer Mode switch with an IPX strapped on top, the two communicating through an internalised Asynchronous Transfer Mode link. The company reckons that an Asynchronous Transfer Mode trunk switch will cost upwards of $75,000 while a multiservice node with Asynchronous Transfer Mode to Frame Relay capabilities – in other words a BPX with an IPX on top – will cost closer to $250,000. There is considerable overlap between the cost of an IPX and a BPX. Cisco’s role in all of this is to work out a way to ensure that Frame Relay-equipped routers will work happily with the Asynchronous Transfer Mode originated traffic. Stratacom’s Chris James says that the the Asynchronous Transfer Mode User-Network Interface is a typical standard in that it has a fair few options to choose from. Cisco, AT&T and Stratacom will be picking a common subset for their own use.
