IP Switching: Battle For The Network High Ground By Art Wittmann   Whether it's stampeding switches, rafts of routers, Ethernet versus Token-Ring debates or shouting matches over the viability of standardizing twisted-pair Ethernet, rarely have networking vendors spread misinformation and FUD--that is, fear, uncertainty and denial--about one another to the extent they are doing so regarding IP switching. To weather the barrage of market bullets, understand that buying an IP-switching solution now means buying a proprietary solution for at least two years. Any other statement about the progress of standards and general acceptance of a particular approach is untruthful or wildly optimistic. That doesn't mean you shouldn't buy into these solutions: They work, can increase the speed of your network and do cost less than buying more routers. Besides, mo st of us tend to buy from one vendor, anyway, so just be aware of the compromise you're making and decide if you can accept it. Cisco Systems is the undisputed king of the networking infrastructure market, but its competitors--among them, Bay Networks, Cabletron Systems, Cascade Communications Corp., Ipsilon Networks, Newbridge Networks and 3Com Corp.--see a glimmer of hope for pulling the rug out from under this 800-pound gorilla. Those taking aim at Cisco are seeking ways to differentiate themselves as they put forth proprietary solutions, which, they hope, will gain enough attention to become de facto standards and, thus, pry off Cisco's market stranglehold, particularly for r outers. What is the problem these vendors want to solve? The answer depends on which vendor answers. All are attempting to make some part of your network go faster for less money. This is an admirable goal. The differentiation comes down to two things: Is the solution intended for wide area or local area, and what compromises will you have to make to realize the benefits? Make no mistake--every vendor requires you to make "minimal and reasonable" compromises in your network, while claiming the other guy's compromises are excessive. Although almost every solution available is IP-specific, the companies involved often, if not always, aim to extend the benefits of their technology to some lesser protocols. IPX is the perennial runner-up, sometimes followed distantly by Apple Computer's AppleTalk, IBM's SNA and Digital Equipment Corp.'s DECNet. Suggest any other protocols, and you're apt to begin hearing whispers of "tunneling" or "partial upgrade." Tactical Ammunition In almost all cas es, you can leave a significant part of your existing enterprise infrastructure in place to handle legacy protocols (anything other than IP), though vendors may not agree with us on this point, particularly when talking about their competition. You'll hear war stories about "truckload upgrades" with the other guy's products--meaning, you 'll need to replace every piece of equipment you own and use if you go with any other vendor's solution. You'll also be warned that the other guys don't fight fair, that the competition isn't interested in standards or is using needlessly complex standards or, even worse, using existing standards in proprietary ways. To a certain degree, such claims contain a grain of truth. But the situation is seldom as dire as these vendors might lead you to believe. Every vendor will tell you how open its standards are, but don't put much weight in such statements. It is going to be a long, long time before there is wide acceptance of standards in this arena; in fact, Multiprotoc ol Over ATM (MPOA) may be the first real IP-switching standard. Cisco and Bay probably will be the first major vendors to support it, along with Newbridge, a champion of MPOA. The wisest course of action is to base your decision on a solid understanding of the enterprise networking problem you are trying to solve, as well as the technological approach taken by the vendors in this market.

State of Middleware: Just Beyond the Limelight by Anthony Frey Updated June 6, 1997