Re: Slam-dunk: email, under 4 seconds, penny/packet

Date view Thread view Subject view Author view

From: Strata Rose Chalup (
Date: Fri Jun 02 2000 - 10:54:56 PDT

Or return-receipt, minus a few degrees of efficiency.

Heck, why don't they just set up a private multicast network
and broadcast each hop to a slightly randomized different group
of "recorder" hosts? Track some unique identifier (like the
message-id, or make the message-id an instance so you can adapt
the system to non-SMTP traffic), run the same type of hashing
checksum on each message, and merely do a store->compare on
the ID/checksum tuple to make sure the complete and accurate
message flashed by. As a safeguard, every now and then, some
station along the way stores the entire message and tells the
hashed branch of the database (Oracle distributed tables, anyone?)
that it has the whole copy Just In Case.

Gateways in and out of the network need to track delivery and
update the entry with the SMTP accept code given by the remote

OK, not a *criminal* waste of bandwidth now, and possibly do-able
with a set of decent VPNs instead of long-lining all across the
doo-dah-day. But jeez. I doubt they're doing it that way.

"The scoops are here! The scoops are here!"


PS- If they had a good enough mesh, there'd only BE one hop
within the mesh, actually...three hops total on their net,
gate-in, mesh, gate-out. That wouldn't suck. Hmm.

"Adam L. Beberg" wrote:
> On Fri, 2 Jun 2000, Rohit Khare wrote:
> > "We've created one of the missing pieces that needs to be in place
> > for the Internet to realize the promise of B2B," Miller said.
> Holy cow chips, they discovered TCP.
> Welcome to the 70's (60's?) Mr. Miller.
> *shakes head*
> - Adam L. Beberg
> Mithral Communications & Design, Inc.
> The Cosm Project -
> -

Strata Rose Chalup []   |
Project Manager                             |     VirtualNet Consulting
iPlanet/Netscape Professional Services      |

Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Fri Jun 02 2000 - 10:56:39 PDT