[24] in Commercialization & Privatization of the Internet
Re: Tardy clarification
daemon@ATHENA.MIT.EDU (jqj@hogg.cc.uoregon.edu)
Mon Oct 22 15:46:16 1990
To: com-priv@psi.com
In-Reply-To: Dave Farber's message of 21-OCT-90 11:47
Date: Mon, 22 Oct 90 12:10:16 -0700
From: jqj@hogg.cc.uoregon.edu
Could someone clarify to me, preferably by private mail since it isn't
directly relevant to this list, why OS12 is "easy" and 2xOS12 is hard
given current technology? Assuming it was possible to implement 1GB/s
data streams on 2 ends of a link, the issue is how to get such a stream
across a network. Seems to me some sharp entrepreneur (like
Bill S, for example) could simply implement an interleaving protocol
where you used several instances of a well known existing protocol
(e.g. TCP) on independent paths to transmit pieces of the data in disk
stripping fashion. That wouldn't be true 1GB/s, but it would be close
enough for the marketeers to argue that they had done something special
and hard.
I think I understand some of the reasons why a true 1GB/s stream built
on routed datagrams is hard. My question is whether that's the right goal,
or whether there are sneaky workarounds that provide a network that we could
sell as "operating at 1GB/s".
