W8SDZ@mit-mc@sri-unix.UUCP (08/20/83)
From: Keith Petersen <W8SDZ@mit-mc> The following file was left on an RCPM system recently. I am forwarding it to stimulate possible discussion on it's desirability. --Keith --forwarded file-- 8/14/83 The national network of Remote CP/M systems is a fantastic resource. Unfortunately, the long-distance communications charges can get very expensive. I live in Hawaii. Even with the lower rates of Long Distance/USA, and calling only after 11 at night or on weekends the toll averages about $10 per hour. What really hurts is the inefficiency of the Christensen protocol for these long-haul connections. Because of the half-duplex nature of the protocol, the delay of the hand-shake/answer-back operation every 128 bytes -- a transfer takes about 30% longer to complete than one made locally. I'm sure there are all kinds of reasons for an overhaul of the Christensen protocol. And I'm also aware of the problems of tampering with a defacto standard. It seems though, that simply by adding an option for 1k "packets" that transfer times could be significantly improved. Over the current 128 byte packets. One-K packets is one possible solution. I would appreciate hearing from others as to other ideas for reducing transmission times. Richard Foulk 1396 St. Louis Drive Honolulu, Hawaii 96816 (808) 735-5013
Chapman.ES@PARC-MAXC.ARPA@sri-unix.UUCP (08/22/83)
I think adding an option to permit larger packet sizes seems a reasonable approach. This would allow the Christensen protocol to remain a standard, while giving the user the option to specify larger packet sizes if he so desires. Cheryl <Chapman.ES @ PARC-MAXC>