On Sep 15, 2010, at 12:16 PM, Richard Bennett wrote: > I think it's actually pretty easy to make the case that a circuit-switched protocol with a sliding window is superior to a stop-and-wait system that required the RFNM from the receiver before every message. In that sense, X.25 was an upgrade over the ARPANET. One problem with coax-based Ethernet was the absence of flow control, which caused bad things to happen to the Internet when IMPs were replaced by Ethernets. Well that's one way of looking at it. Others might draw different conclusions. This discussion reminds me of the discussion at BBN of X.25 style packet switching vs. the Internet datagram approach. I think the outcome was clear. Bob > > RB > > On 9/15/2010 12:04 PM, Bob Hinden wrote: >> On Sep 14, 2010, at 5:08 PM, Richard Bennett wrote: >> >>> I wonder how many people realize that X.25 was a direct descendant of ARPANET, and that BB&N became a leading supplier of X.25 hardware simply by continuing the IMP down its evolutionary path. >> I was at BBN at the time this was going on. BBN implemented X.25 because it needed a "standardized" interface to the network instead of BBN's proprietary 1822 interface and choose X.25. X.25 was developed in parallel to the Arpanet and I disagree that it "was a direct descendant of ARPANET". It has a very different interface (connection oriented vs. message oriented) that IMHO was not an improvement. >> >> Bob >> >> p.s. I suggest that BBN use Ethernet instead but that didn't get any traction. I am pretty sure the world would be different had they followed my suggestion. >> >> > > -- > Richard Bennett > _______________________________________________ Ietf mailing list Ietf@xxxxxxxx https://www.ietf.org/mailman/listinfo/ietf