home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.dcom.modems
- Path: sparky!uunet!stanford.edu!ames!sgi!rhyolite!vjs
- From: vjs@rhyolite.wpd.sgi.com (Vernon Schryver)
- Subject: Re: Latency Question
- Message-ID: <pea6luk@rhyolite.wpd.sgi.com>
- Organization: Silicon Graphics, Inc. Mountain View, CA
- References: <1992Sep2.034404.18553@wdl.loral.com> <BOB.92Sep4104913@volitans.MorningStar.Com>
- Date: Fri, 4 Sep 1992 16:13:31 GMT
- Lines: 53
-
- In article <BOB.92Sep4104913@volitans.MorningStar.Com>, bob@MorningStar.Com (Bob Sutterfield) writes:
- > In article <1992Sep3.003250.14571@zip.eecs.umich.edu> dmuntz@quip.eecs.umich.edu (Daniel A Muntz) writes:
- > Latency hasn't changed at all between my WB's after upgrading to
- > 5.01. Ping settles to ~200ms (56 data bytes).
- >
- > That packet is so big (~88 octets, counting ICMP headers) that it's a
- > better measure of in-modem data compression (V.42bis) than of
- > modem-induced latency, or of the responsiveness that the user feels
- > when typing in a telnet session that crosses the link.
- >
- > To test modem-induced latency, you want the smallest packet possible.
- > ...
-
- > These interactive latencies are far better than what you see with ICMP
- > pings, and they better represent the type of latency that matters:
- > what a user feels. It's encouraging that they're generally below the
- > magical 200ms perceptibility threshhold.
- > ...
-
- That's true if you want to measure the latency associated with
- keystroke echos during dumb-terminal telnet sessions.
-
- If you want to measure the latency of keystroke echos with xterm, you
- want to use something with the much larger than 1-byte of payload. X
- is not like telnet. Non-host-compressed pings with a small "-s size"
- are probably a lot closer to what you'll see with xterm.
-
- If you really care about typing latency latency, the best thing to do
- is get rid of all of that telnet/TCP/IP/{PPP,SLIP} nonsense, and use a
- simple, dumb-terminal protocol directly over the modem.
-
-
- It seems to me that the premise that single data-byte latency is all
- that matters with telnet is wrong. I use `rlogin` over compressing
- SLIP and v.32bis/v.42/v.42bis a lot. For for many hours/day and many
- days/week. The single character latency is not what really hurts,
- whether it is 10 or 100 ms. What is noticable is the latency to `vi`
- edits or `ls`, when the answer is 1KBytes or even 100Bytes.
-
- FDX character echo times have been perfectly fine for 25 years, since
- the days of 110 baud 10 Bytes/sec FSK 103/113's, except for short-lived
- oddities like putting 42 byte rlogin/TCP/IP/SLIP packets over 2400
- bit/sec modems. (That particular scenario was the genisis of cslip.)
- Compare reading netnews over a modem with looking at the local disk.
- The latency that matters is not the time to echo characters, but the
- time to respond to a command.
-
- Finally, anyone concerned with character echos over telnet who cannot
- dump the TCP stuff should should use the Cray inspired telnet line
- modes.
-
-
- Vernon Schryver, vjs@sgi.com
-