Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 15 Dec 2002 12:44:54 +1030
From:      Greg 'groggy' Lehey <grog@FreeBSD.org>
To:        Bsd Neophyte <bsdneophyte@yahoo.com>
Cc:        FreeBSD Questions <questions@FreeBSD.org>
Subject:   Re: Hubs and switches (was: uninformed qstn...)
Message-ID:  <20021215021454.GA53628@wantadilla.lemis.com>
In-Reply-To: <20021215020733.46093.qmail@web20106.mail.yahoo.com>
References:  <20021215012545.GB144@wantadilla.lemis.com> <20021215020733.46093.qmail@web20106.mail.yahoo.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Saturday, 14 December 2002 at 18:07:33 -0800, Bsd Neophyte wrote:
>
> --- Greg 'groggy' Lehey <grog@FreeBSD.org> wrote:
>> I'm not sure you meant to write what you did, but I'm not 100% I
>> understand.  But yes, latency isn't an issue.  Even if it is, switches
>> still win.
>
> My point is that generic switches are not faster than hubs, because of the
> added latency associated with the store-and-forward method.  Cisco 2900's
> have the ability to do both types of switching.
>
>> I tried a test here between two machines on my network.  In each case,
>> the data went via the Cisco 2900 switch and then either via a hub or a
>> second switch.  The remote machine has a 10 Mb/s interface.  Here are
>> the results (average ping time):
>>
>> Switch, normal ping:        0.756 ms
>> Hub, normal ping:	    0.744 ms
>>
>> Switch, 1500 bytes:	    4.251 ms
>> Hub, 1500 bytes:            4.004 ms
>>
>> Switch, 1500 bytes, load:   4.244 ms
>> Hub, 1500 bytes, load:      4.513 ms
>>
>> The "load" was a single concurrent tar over the network.  I must say
>> I'm impressed how little degradation there was, but it's clear that
>> the latency savings on a hub are more than offset by its performance
>> under load.
>
> I would offer the rebuttal that when you look at costs, having a hub
> deal with 12-16 nodes would more than suffice.

For some definition of "suffice".  But why?

> For example, a new catalyst 2912, depending on options, will run you
> between $2000-$5200, while a 12 port fast ethernet hub will run you
> about $200-$300.

If you spend that much money on a hub, you're really wasting it.  As
discussed on this thread, you can buy 16 port switches for under $100.

> BTW, I don't know if your switch is configured for cut-through
> switching or store-and-forward.  If it is set for store-and-forward,
> I'd suggest you change it to cut-through.

I don't care.  We're looking at a difference of 200 µs here.

> I suppose if there isn't a big price difference between a generic
> switch and a hub, the switch would seem to be a better bet.

There isn't, and that's the point we've all been making.

Greg
--
When replying to this message, please copy the original recipients.
If you don't, I may ignore the reply or reply to the original recipients.
For more information, see http://www.lemis.com/questions.html
See complete headers for address and phone numbers

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20021215021454.GA53628>