Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 18 Mar 1996 20:56:53 -0500
From:      "Louis A. Mamakos" <louie@TransSys.COM>
To:        Terry Lambert <terry@lambert.org>
Cc:        alk@Think.COM, hackers@freefall.freebsd.org
Subject:   Re: hackers-digest V1 #986 
Message-ID:  <199603190156.UAA13663@wa3ymh.transsys.com>
In-Reply-To: Your message of "Mon, 18 Mar 1996 10:46:09 MST." <199603181746.KAA21733@phaeton.artisoft.com> 

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

> > Of course, this is orthoginal to terminating T1 leased lines from
> > customer locations; some of them want their very own port from their
> > premise to "mine", which isn't "shared".  You can fix part of the
> > problem space, but not all with one solution.  Thus, the extreme
> > interest in directly terminating M13 framed DS3 circuits carrying 28
> > T1 circuits.
> 
> You have to have two clouds.  You sell overcommit in the first, and
> you don't overcommit the second.  At a 50% load, it costs on the order
> of twice as much for the committed rate.  This assumes all common
> costs (hardware, installation, lines, etc.) have been subtracted
> out.

The problem isn't a technical one; it's a marketing and product
packaging one.  It's like trying to argue over what color it should be :-)

> 8-).  I bet *against* ATM.  Leaky-bucket is OK for voice, and can be
> OK for video, assuming Cell animation with scene refresh.  But it rots
> out for data, which *must* get there *eventually*.  I have yet to see
> a working "source quench" without specialized hardware... besides,
> with the number of audit records that would have be generated, they'd
> be in the same boat as FR: no way to bill connect time.  Not that that
> is a bad thing.  8-).

ATM may be OK in the local loop as a fine-grain muxing scheme, so long
as you don't over commit the facilities.  Of course, the problem is to
have the LEC's actually offer ATM as a service...  On the other hand,
DS3 Frame Relay seems to work Just Fine..

> > Sorry for the diversion; this is pretty far afield from FreeBSD, 'cept
> > that I run it on my machine and it works pretty good.
> 
> No it isn't; it bears on what is a profitable use of time for drivers,
> protocol stacks, and the ability to use FreeBSD for ISP services based
> on communications throughput.  Is UUNET still using those Sequent MP
> boxes?

Not for years; certainly not since I've been working there (about 2.5
years now).  It's been a combination of Sparc and Pentium/BSDI
platforms.  There's a bunch of 'em these days.  I think we've got on
the order of 20 or 25 machines doing NNTP feeds at the moment.  The
Pentium machines are mostly CDDI attached, and we *sustain* about 30
Mb/s of just NNTP news feeds.  High pressure USENET sewage pumping
station..

louie






Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199603190156.UAA13663>