[1417] in Commercialization & Privatization of the Internet
Re: Volume-sensitive charging
daemon@ATHENA.MIT.EDU (Martin Lee Schoffstall)
Sun Sep 29 18:52:20 1991
To: SEAN@sdg.dra.com (Sean Donelan)
Cc: com-priv@psi.com
In-Reply-To: Your message of "Sun, 29 Sep 91 16:55:29 CDT."
Date: Sun, 29 Sep 91 18:49:03 -0400
From: "Martin Lee Schoffstall" <schoff@psi.com>
>Some service providers such as PSI put contractural guarantees on
>certain aspects of internetworking, provide backup capabilities etc..
Does this mean PSI will guarantee reachability of a site even if it
is on the other side of an ANS, Alternet, or NEARnet (etc...) router?
Now that would be exciting. But can you expect to be able to call
MERIT because you can't reach a PSI customer, after all they aren't
getting money to support your customers. And likewise when an ANS
router takes a dump, do you expect PSI to provide reachability to
ANS customers? And who is going to do something about the delays
to sites on CA*net (or TNnet, etc)?
The answer is no. That is why I said "certain aspects". Ten years into
this, as an engineer, you focus on what are the principal problems -
one principal problem are the single points of failure of WAN
interconnects, ie the local loop and terminus router. Provide a
connectivity solution to this problem, and guarantee that the service
is restored through backups and a whole class of reachability issues
go away. [ANSNet is bad example since its backbone is probably an
order of magnitude worse than the worst regional network due to the
equipment that they committed to use - RS6000/Unix routers]. Co-located
commercial quality routers such as in PSINet/JVNCNet/Infonet are probably
two orders of magnitude more reliable as networks.
>However, many organizations are only interested in how many bps/$ they
>get, and all other issues aren't even considered.
True, but they also expect/assume that the bps/$ will work end-to-end
a high percentage (> 99%) of the time. Paying $$$ for bps to nowhere
isn't very cost effective. People care, but the internetworking industry
is still young enough that people haven't gone through a second buying
cycle.
This is the opening that the "big boys" want to use to go back to one
network and remove most of the current players. I've seen it used time
and time again. Of course to date they have "proved" the opposite, but
it really doesn't matter, if you have good audio visual materials, have
all the right initials, etc. reality doesn't have to intrude.. at least
in the short term.
marty
