Conference 7.286::fddi

634.0. "cisco/FDDI/routing/bridging" by BOWLES::BOWLES (Bob Bowles - T&N EIC/Engineering) Sat Jul 04 1992 16:46

    
    Apologies if some of these questions seem obvious, but I must be sure
    of these observations in a sensitive customer environment.  And I'm
    relatively new to cisco hands-on experience...
    
    Customer has implemented an extensive cisco based LAN/WAN with
    multi-protocol routing and bridging.
    
    At the central data center, customer has implemented 3 or 4 separate
    ethernets via combinations of Synoptics fiber and UTP hubs.  Bridging
    and routing between these Ethernets is accomplished by cisco AGS+ boxes
    with the CSC3 processors.
    
    Now the FDDI twist ... using a synoptics FDDI hub, customer has put
    FDDI controllers in the 3 AGS+ boxes and is planning to migrate some
    high-end VAXes onto the ring.
    
    Today, the 6xxx class VAXes are all attached to various of the
    Ethernets in the building via XMI enet adapters.  I come along and as
    part of a network evaluation, I begin sending large amounts of data
    between the VAXes with my favorite traffic tool DTSEND.
    
    Between two VAXes on the same ethernet, I am routinely getting 700-947K
    bytes per second.  DTSEND claims this is around 6-7.5 Mbaud, btw.
    Now, between two identical class VAXes on separate ethernets,
    traversing the ethernets via AGS+ <-> FDDI <-> AGS+, I can only get
    about 90K-140K bytes per second (0.7 Mbaud - 1.1 Mbaud).
    
    The cisco boxes are routing DECnet, and my tests use packets sizes of
    540 bytes and 1400 bytes.
    
    So we have a bit of router overhead/limit going across the FDDI with
    the cisco routers.  I am having a little trouble understanding why
    routing onto an FDDI is of any advantage to full-rate bridging when the
    routers are restricting traffic so much.  All ethernets in the building
    are around 5-10% utilization.
    
    When the customer moves the VAXes onto the FDDI ring, I am concerned
    about LAT bridging.  They are already bridging LAT all over the WAN in
    addition to using some Ethernet based AppleTalk/LAT gateways to allow
    AppleTalk only users to make LAT connections into the VAXes.  Are they
    going to be adding to the LAT delays by bridging LAT through these
    cisco boxes with obviously limited throughput?
    
    Does cisco even do translation-bridging yet?  I don't see it explicitly
    mentioned.  Encapsulation-only would make most of my questions moot.
    
    Why shouldn't this customer use full-rate FDDI-Ethernet bridges from
    Digital?  Obviously there will be some minor routing changes required
    with the AppleTalk and TCP/IP protocols on the Ethernet segments if
    they move to bridging, but DECnet is easy with the single area already
    in use.
    
    Customer is planning to upgrade the cisco boxes to the next processor
    level (CSC4 I believe), but I really have my doubts whether they will
    be able to give the customer any worthwhile use of the FDDI
    configuration.
    
    Any comments are most welcome.  Offline discussion of who this customer
    is and why our recommendations are so important is also welcome.
    
    Bob (onsite for another week)

    
    Some answers from cisco:
    
    FDDI bridging is still 'transit' (i.e. encapsulation) between Ethernets
    (That rules out LAT and NI traffic)
    
    My DECnet throughput tests should be seeing much higher numbers routing
    over the FDDI between two ciscos.  They suggested tinkering with 'fast
    swtiching' ON or OFF to determine any differences ...
    
    Sounds like cisco still has some issues to resolve around FDDI.
    
    

    Bob,
    
    I just finished a design using a bunch of cisco Bridge/routers that are
    tied into FDDI. The real story (be VERY careful about believing cisco
    on ANYTHING) is that the CSC-FCI card is in fact encapsulation
    bridging. There is supposed to be a new card, CSC-FCIT which will
    aledgedly ship in September that will do full translation bridging.
    There is of course, an upgrade to the bridging software that goes along
    with the new card. By the way, don't forget that either card requires
    the use of one od cisco's so called "applique" packages. THis is a
    bunch of connectors and cables etc. that cisco charges $10,000 for (yea
    thats right $10K !). The applique part # is APP-LMM.
    
    	Regards,
    
    		John DTN 328-3090 

    Are you saying that when we bid against cisco for FDDI environments,
    each and every one of their FDDI cards will require that the customer
    purchase a mandatory package for $10,000?
    
    If you add this to the prices of the "known" cisco pieces, it would
    seem that they are a great marketing firm?  Please verify if the above
    statement is true.
    
    If so, we need to let the field and our perspective customers know
    about this so we don't continue to get the short end of the stick so
    often.
    
    Regards,
    
    Mark
    

From the cisco price book

For AGS+

CSC-FCI FDDI interface                 				$4000
interface connects to the cBus (req a fiber optic tranceiver 
applique for dual connect FDDI ring on multimode fiber, APP-LM1 or APP-LM2)

Applique assemblies

APP-LM1 One fiber optic tranceiver applique for dual attachment FDDI $10000
APP-LM2 Two fiber optic tranceiver applique for dual attachment FDDI $20000

Then if the customer don't have, he needs the cBus controller

CSC-CCTL cBUS controller                                           $2200

AND bridging software comes extra 

SW-BSA-GS Bridging Software 					$1800


Now my private comments

MINUS some heavy discounts ofcourse....


NOTE - it is not documented but if the were to connect as SAS they 
might be able to use one applique for two FDDI adapter cards...

They are about to come out with a more performant FDDI card...
and soon they will master translation bridging...