marc@dumbcat.sf.ca.us (Marco S Hyman) (05/05/91)
In article <3821@sixhub.UUCP> davidsen@sixhub.UUCP (bill davidsen) writes: > Now, on the XOFF, any system with a FIFO may well keep right on > sending after the XOFF, and I think there's a standard which calls for > XOFF 1 sec before buffer full, I just can't remember where I saw it. We > hit this in 1978 or so when running a mainframe into an old S100 system > at 38.4. Time for the back of the envelope. At 38.4 Kbps speaking async 1 second is ~ 3840 characters. That's a lot of buffer space destined to never be used. Last week, using a 256 byte buffer I had to change the "flow off" point from 2/3 full to 1/2 full when speaking to a 3b2/600 at 19.2 Kbps. As the 3b2 got busy I was dropping characters -- about 10-20 implying the buffer should have been 80-100 bytes or about 50 milliseconds worth. Going to 64 milliseconds worth of buffering did the job no matter how bad the load. Note: a VAX 780 running VMS (ugh!) had no problems shutting down within the original 32 millisecond of buffering. -- // marc // home: marc@dumbcat.sf.ca.us pacbell!dumbcat!marc // work: marc@ascend.com uunet!aria!marc