[NTLUG:Discuss] Router or switch?
Joel Sinor
jsinor at comcast.net
Mon May 5 17:36:36 CDT 2003
On Mon, 05 May 2003 16:55:39 -0500
"Darin W. Smith" <darin_ext at darinsmith.net> wrote:
> On Mon, 05 May 2003 14:25:24 -0500, Greg Edwards <greg at nas-inet.com>
> wrote:
<snip switch discussion>
> I would recommend a switch. The only two useful features I'd look
for
> on one beyond being a 100BaseTX switch are:
> 1) Ability to run full duplex
> 2) Auto-MDIX
>
> Auto-MDIX is a great thing! When I wired up my house last year, I
> used cheapo Fry's palm-sized switches that are full-duplex capable and
> support auto-MDIX. I just plugged everything together and was done.
> No more "do I need a crossover cable?" It just figures it out, and
> does it a heck of a lot faster than I can think "I only need a
> crossover cable if I'm going NIC-to-NIC or switch-to-switch
> (non-uplink)".
>
> --
> D!
> Darin W. Smith
> AIM: JediGrover
>
> "For want of a nail, the shoe was lost:
> For want of the shoe, the horse was lost;
> For want of the horse, the rider was lost;
> For want of the rider, the battle was lost;
> For want of the battle, the kingdom was lost,
> And all for the want of a nail."
>
> The moral of the story: details matter.
>
> _______________________________________________
> https://ntlug.org/mailman/listinfo/discuss
Yes, now that I know about auto-MDIX I am convinced it is The Way To Go.
Now, as to the rest of this, I believe you are right about the way
traffic is sent to ports. However, one of the things I have run into as
a justification for higher-cost switches (in addition to the management
and higher-level protocol handling you mention ) is bandwidth
sharing/nonsharing. I still don't understand the specifics as much as I
would like, as I am far from an EE, but essentially every switch has a
backplane upon which all traffic travels. The individual conections are
indeed made between individual ports(port a is connected to port f, and
traffic only"travels" that circuit) but there is only so much wire for
all the talking to occur on. High end switches' market blurbs include
things line "#GB backplane" and "#GB dedicated bandwith to each port"
the cheaper switches, like the one I mentioned, would likely therefore
be soundly trounced in performance comparisons with a high-end Cisco or
3com switch. Of course, as with all things Computing, this really
depends heavily on the application. The small cheap switches are meant
for SOHO use, and for most people are more than adequate. The high end
stuff is for managing enterprises, etc. Of course, "most people" does
not include those with large beowulf clusters running a distributed
application cracking RSA or modelling nuclear explosions. The less
extreme example would be if you have many users connected to a given
switch constantly sending large files across simultaneously. My network
traffic does not look like that yet at home, so YMMV. ;)
More information about the Discuss
mailing list