kwe@buit13.bu.edu (Kent England) (03/05/90)
In article <21184@bcsaic.UUCP> carroll@bcsaic.UUCP (Jeff Carroll) writes: > Related question: how does FDDI compare, cost-wise, to HPPI? > As I understand it, HPPI has a 75 meter limit on coax media. I am not aware of any other media options. (I am sure someone will enlighten us if there are options.) Remember, HPPI is warmed-over HSC, a mainframe-to-mainframe interconnect standard. While FDDI started out as a machine room interoperable interconnect standard, it left those kind of media and signaling limitations behind when it decided to follow Pronet-80. FDDI is limited to fiber, which means that you can't substitute cheaper media and trade-off distance and stay within the spec. (That is one of the things I like about Pronet-80; you have some media choices when the fiber optic interface is outboard of the node.) Therefore, there is no comparison between FDDI and HPPI except in the machine room definition of "local area". HPPI is extremely local (like a bus) and FDDI is nearly a WAN (will be a WAN when someone does it). But, I imagine that if FDDI is not already cheaper, it will be soon. Remember, the cost of a proprietary interface is directly related to the cost of the machine being interfaced. (England's 12th law of successful marketing to captive communities.) The extension to interoperable standard interfaces is slightly modified to include a deflation factor based on the cost of the cheapest machine with a standard interface. So, FDDI wins at the moment. :-) Kent England, Boston University
schoff@psi.com (Marty Schoffstall) (03/06/90)
Just to add another wrinkle to the discussion, FDDI is complex, many vendors (specifically in the area of the station management) do things in software and this will continue. HPPI is a rather simple all hardware solvable mechanism. Lastly I have seen plans for fiber between HPPI's. -- Martin L. Schoffstall Performance Systems Internationl Inc. Reston Virgina, US schoff@psi.com
vjs@rhyolite.wpd.sgi.com (Vernon Schryver) (03/06/90)
In article <53312@bu.edu.bu.edu>, kwe@buit13.bu.edu (Kent England) writes: > ... FDDI is > limited to fiber, which means that you can't substitute cheaper media > and trade-off distance and stay within the spec. ... What would be wrong with a pair (or more) of cooperating stations having copper, string or any other media between them, and fiber to their neighbors? There is no way I can see for the token to see what kind of media it is passing thru. Of course you would not be using officially sanctioned MIC connectors between the cooperating stations, and the Protocol Police might come and take you way. What if you had a dual-attach "concentrator" with 0 M-ports and 20 MAC's connected with copper, whether on a back plane or in coax, and what if this "concentrator" happens to have 20 fast CPU's with disks, operating system, terminals, and so forth, and further, happens to put many normal LLC frames on the rings? What if this "concentrator" happens to be a physically large machine, spread over an entire building? It would be completely within the letter if not the spirit of FDDI spec., so you would be safe from the Protocol Police. This is not a pure academic nit, given the current cost of the optics. (Yes, as you might guess from my signature, I've heard the "trust me, they will be real cheap real soon now" pitches from the ODL vendors. As you might also guess, I currently have no choice but belief.) Vernon Schryver Silicon Graphics vjs@sgi.com