Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 13 Feb 1998 10:41:54 -0500 (EST)
From:      "John T. Farmer" <jfarmer@goldsword.com>
To:        freebsd-isp@FreeBSD.ORG, joe@thebestisp.com
Cc:        jfarmer@goldsword.com
Subject:   Re: Fw: FreeBSD firewall questions
Message-ID:  <199802131541.KAA09287@sabre.goldsword.com>

next in thread | raw e-mail | index | archive | help

On Fri, 13 Feb 1998 00:12:08 -0600 "Joe" said:
>OK... For starters I am not on this group to knock FreeBSD that would be
>STUPID. I use FreeBSD and that is why I am on here. As for the hub things  I
>concead defeat and am going to try some more testing. The thing is that I
>have 3com and SMC hubs here along with a few assorted others and have always
>had them top out at about 6Mbps with every card I have tried (and yes the
>cables are wired correctly) so at any rate I was not trying to start
>anything and I sincerely thought that the information I was giving was
>correct. Also several months ago I spoke with a "tech" at 3com about there
>officeconnect 10Mbps hub and that is initialy where this theory (the 60%)
>began for me. But while I am on this what is the "best" nic?

No offense taken :^>

Actually, after I went to bed, I realized that there is one mode in
where a cross-over cable would increase performance.  However, it is 
dependant on the capability of the two devices to "max-out" the 
number of packets/second & the packet sizes.

Twisted-pair Ethernet was based on the physical configuration of
10base10 where the collision-detect & jabber circuity is located in the 
transceiver _at_the_baseband wire.  (that's what the AUI port is for,
cabling from device to transceiver).  In twisted-pair, circuity for
collision & jabber detect is _in_the_hub_, so in theory, the two 
devices with a crossover cable would never see collisions.

This is how full-duplex is/was added easily to 10baseT, there are 2
channels between the device and the "collision domain or space."  
In the special case of a cross-over cable, the collision space
collapses to null.  In theory, the maximun throughput would then
be limited only by the Ethernet specifications for packet size &
the required inter-packet "gap."

Note: this is only in theory, and in any case, the gain would be
on the order of 0.001%

Regarding your 6Mbps throughput, I would check into the packet sizes
and what the actuall maximum throughput of the NICs you were using.
(Note: I _have_ in the past been able pump 10Mbps through Ethernet
using a dedicated PC (DOS & special s/w).

As to your other questions, I like the Intel cards for FreeBSD systems.
I've also used 3com 3C5xx cards in several types of systems with good
success.

John

-------------------------------------------------------------------------
John T. Farmer			Proprietor, GoldSword Systems
jfarmer@goldsword.com		Public Internet Access in East Tennessee
dial-in (423)470-9953		for info, e-mail to info@goldsword.com
	Network Design, Internet Services & Servers, Consulting

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?199802131541.KAA09287>