Date: Sat, 16 Nov 1996 14:53:29 -0500 From: Andrew Herdman <andrew@why.whine.com> To: Bradley Dunn <bradley@dunn.org> Cc: dennis <dennis@etinc.com>, isp@freebsd.org Subject: Re: changed to: Frac T3? Message-ID: <Pine.BSF.3.95.961116145209.451A-100000@why> In-Reply-To: <Pine.WNT.3.95.961116124539.-512999s-100000@swoosh.dunn.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, 16 Nov 1996, Bradley Dunn wrote: > On Sat, 16 Nov 1996, dennis wrote: > > > >On Fri, 15 Nov 1996, dennis wrote: > > > > > >> What I was saying was that I dont thing unix can route a steady > > >> 86Mbs data stream, so a full T3 on a unix box may very well be > > >> overkill. > > > > > >Hmmm...Apparently you are not aware of the Ascend GRF 400. > > >http://www.ascend.com/products/grf400/grf400index.html > > > > Perhaps you haven't read it yourself? They are certainly not running anything > > similar to standard unix....they "cheat" by maintaining on-board caches so > > packets don't have to pass through the IP layer, as BSD design requires. > > Certainly you can do something similar for BSD systems, but it won't > > be a standard release O/S afterwards. Such things are OK if you are building > > a special-function system, but non highly desireable for general purpose O/Ss > > Exactly, but you seemed to be saying that unix could not route at that > speed. The Ascend embedded OS is a hacked unix. It uses gated, but you > could in fact use anything that writes to the unix routing socket. I call > that unix. > > -BD > The UNIX itself doesn't actually do the routing. It creates the table and download's it down to the actual interface cards which use a type of silicon switching to route the packets. The UNIX portion of the box simply makes the tables. Andrew
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.3.95.961116145209.451A-100000>