Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 12 Feb 1998 09:00:15 -0800
From:      Don Lewis <Don.Lewis@tsc.tdk.com>
To:        Benedikt Stockebrand <benedikt@devnull.ruhr.de>, "Joe" <joe@thebestisp.com>
Cc:        <freebsd-isp@FreeBSD.ORG>
Subject:   Re: Fw: FreeBSD firewall questions
Message-ID:  <199802121700.JAA20432@salsa.gv.tsc.tdk.com>
In-Reply-To: Benedikt Stockebrand <benedikt@devnull.ruhr.de> "Re: Fw: FreeBSD firewall questions" (Feb 12,  4:48pm)

next in thread | previous in thread | raw e-mail | index | archive | help
On Feb 12,  4:48pm, Benedikt Stockebrand wrote:
} Subject: Re: Fw: FreeBSD firewall questions

} As far as latency goes: How much latency is caused by a hub (opposed
} to a switch, which is too expensive anyway) and will it be noticeable
} if all data is subsequently sent across a long distance connection?

>From my reading of the 802.3 repeater spec, the maximum delay through
a 100Mbit Class II repeater is 84 bit times.  Add about 25 for a
Class I repeater.  I don't have a copy of the 10Mbit spec, but I would
expect the number of bit times to be quite a bit less.

I'd expect that using a hub would cause a slight increase in the
collision rate (depending on how well the NIC and driver can send
back to back packets with the minimum interpacket gap, and the
overall traffic pattern), and a slight increase in the size of the
runt frames generated by collisions.

If adding a hub causes a big change in throughput, one of the first
places I'd look is at the NICs.  To really track down the problem,
you really need to examine the bits they fly past ...

			---  Truck

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



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199802121700.JAA20432>