| This question having not been answered, could I ad my vote for some more
information on the subject of Transit Delays within the FDDI bridges.
The subject of Bridging verses Routing within a LAN is common thesedays
where our FDDI offerings are being competed against by multiport Bridge/Routers
such as those made by Cisco and Wellfleet. I believe transit delay is one
area where we can show real performance benefits.
Is it therefore possible to give an idea of the ballpark we are talking about
for the time to process a packet (ie make a forward/filter decision, carry
out the translation and place the packet onto an "output buffer", I realise
that factors such are token latency (in the case of FDDI) and outbound
congestion (in the case of Ethernet) will have very significant impact on the
actual time taken but at least it shows how the bridges perform under ideal
conditions. I believe the BRouter hub products will loose out heavily here as
their transit delays will be a bottleneck even in light traffic.
I realise that the information is not published owing to its potential for
misuse, however if explained in the correct way it could be a very powerful
argument for our FDDI solutions, so any information will be of assistance.
If information provided is confidential and not to be disclosed, please state
that, but it will still be useful to have it.
Thanks in advance for any help.
Dave Swanson
London - UK
|
|
Transit delay is normally a function of implementation and HW speed. There is
no inherent reason why a BRouter should have a higher transit delay than a
bridge for a given input traffic load. There are certainly implementation
reasons why this may be so, but it's not guaranteed to be true by any means.
The DECNIS 600 is a BRouter, including full transparent bridging (not Cisco's
NI-FDDI-FDDI-NI hackette) when we support FDDI. If you produced your transit
delay argument against Cisco etc. now (even assuming that it's valid, which
I doubt) then what will you do when the DECNIS is available ?
Eric
|
| >> <<< Note 145.2 by MARVIN::DAVISON "Eric Davison" >>>
>> -< Sounds dangerous ? >-
>>
>> If you produced your transit delay argument against Cisco etc. now
>> (even assuming that it's valid, which I doubt) then what will you do
>> when the DECNIS is available ?
Eric,
I was not proposing to go to the customer making sweeping
generalisations about BRouters vs. FDDI bridges in terms of
performance, but I am aware that the customer is carrying out these
studies on a theoretical basis. I also understand (though I might be
wrong) that Cisco (as opposed to BRouters in General) show up quite
badly on transit delay under heavy load as so by providing the
information as indicvated in .1 this would be source for comparision.
However since investigating this notes file further it appears that
things such as token acquisition time and outbound congestion which can
be in the order of milliseconds are likely to swamp this figure (which
I guess is in the tens of microseconds region).
Dave
|