[ih] Design choices in SMTP

[Dave Crocker]

On 2/7/2023 8:22 AM, Craig Partridge via Internet-history wrote:
> And just because this is a history list, and capturing stories like this
> are useful.

As I recall, I had little or no contact with the SMTP pipelining effort, 
other than joining in the general community support of the effort. (One 
took a position against the views of Ned, Van or Phil at one's peril.  
Not as a matter of their personalities but their irritatingly consistent 
tendency to be right.)

Also, I'd learned about the miracle of pipelining around 1981, with 
MMDF.  Another grad student, Ed Szurkowski, had done the link-level 
protocol that MMDF used over telephone connections. (30-120 Bps, back 
then...).  It was a lock-step exchange, essentially half-duplex.  As our 
base of client sites for CSNet increased, the wasted bandwidth 
eventually became intolerable.

Looking for a cheap way to do better, I modified his code to allow 2 
outstanding packets rather than one.  The result was that the modem 
transmit light was pinned on, with no loss of reliability or functionality.

d/

-- 
Dave Crocker
Brandenburg InternetWorking
bbiw.net
mast:@dcrocker at mastodon.social