VanLUG Email Archive
dlinford@vcn.bc.ca
Tue, 22 Feb 2000 15:19:03 -0800
in <200002220842.AAA00907@sheridan.sibble.com>,
"Harondel J. Sibble" <help@pdscc.com> writes:
>On 21 Feb 00, at 17:08, dlinford@vcn.bc.ca wrote:
>> On the other hand, it increases availability of the "logica bus",
>> which can dramaticaly decrease effective latency. All depends on
>> usage/environment. It'd be nice to know the technical argument
>> against switches that's be put forward for you application.
>
>basically for real work applications, the switch will offer better network
^^^^ <== you spelt "most" wrong?
>performance period.
That plain biggoted and _wrong_. It is possible to build switches
that would be on par, for this pathological case. Care to explain
the design choices the the audience? For the cases where this _may_
be a problem (I doubt it would be, but I'm not the one that was
making the claim - but I did ask to hear the argument), the one
design choice _always_ adds latency. It ain't difficult to draw
the graph, for the pathological case, and see that there is no
way to get 'roud (i.e. transforms) the _potential_ problem.
Just 'cause it doesn't effect you, don't make it not "real".
> . I can't think of any circumstance where a hub will out
>perform a switch...
That's nice, you can't think of one. You may wish to read up some,
and try to figure out what is meant by sequential dependency.
Some groups that couldn't, it seems: TVI, iStar, MotionWorks,
Praxis Technical Group, Helix, Gandalf. Wonder how Praxis is
doing (seems they don't talk about Hesse now), I know enough
of the fate of the other fools.
Some of computing involves engineering. Hard to notice now adays...
dean
-- This message came to you via the Vancouver Linux Users Group mailing list. For unsubscription instructions do not email the list, but rather send mail to <vanlug-request@gweep.bc.ca>.
This archive was generated by hypermail 2.0b3 on Tue 22 Feb 2000 - 23:20:48