[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

PPP & Compression: Useful or Useless?

Leonard Monk wrote:
> Another concentric client (Jay Hofacker, jay@tachyon-inc.com)
> suggested removing the bwd_comp module, and this seems to work.
> I suppose there is some loss of performance by removing this
> compression device.  If you have a suggestion for a better
> solution, please let me know.

If you were getting dropped a few seconds after connecting, then
it's likely that this compression protocol could not be negotiated
anyway. Ergo your performance can't be considered to "suffer". Note
that compression protocols, while improving bandwidth, will always
introduce some amount of latency since encoding/decoding takes
compute time. Now don't everybody jump on me to tell me this amount
of time will be unnoticeable. I wouldn't be surprised if it is
effectively unmeasureable. The fact remains that it exists, and this
is the point I'm trying to make. 

To continue, consider what kinds of things take a long time to 
download. That is to say, ask yourself what sort of data you are 
downloading that will take a lot of time. I assert that most of the
time it'll be either a jpg or gif (WWW) or something which has been
gzipped. Guess what: you're not going to get much compression of these
suckers because they're already compressed. If you don't believe me 
then try some empirical tests yourself. This is because basically
all lossless compression algorithms take advantage of repeating patterns
and express these patterns with less bits. Once you've done this 
operation with a good compression algorithm, you've got data that's
getting pretty close to random, and you'll be hard pressed to compress
it further. Try gzipping a .gif file sometime and you'll see what I

The long and the short of it is that I don't believe there's a big
bang to be gotten from compression over PPP for most people's usage
patterns. That's why I don't bother with it.

> I appreciate all the help people are willing to give, especially
> on this mailing list.  Still, I wonder if it is not becoming a
> critical problem for Linux that so many potential new users have
> such a hard time overcoming the miscellaneous obstacles to getting
> a good PPP conection.  I don't believe there is an easy fix, especially
> when one has to deal with ISP's who are very uninformative.

Yes, I believe it's a bigger problem than some Debian folks think it
is. I get the feeling that many Debianers (those making packages, 
running ftp/web sites, you know the ones) seem to have this idea that
most people are going to buy a Debian CD. I personally hold the belief
that *most* people get Debian via ftp, which generally presupposed 
a PPP connection. I believe the trouble could be solved by writing
a very intelligent "Wizard-like" (if I may use this M$ paradigm as
a basis for comparison) program which would step by step go through.
Such a program would actually need to *itself* dial the phone and
scan/interpret what it was getting back. Such a program wouldn't be
all that complex to create but would take some time. I'd like to do
it myself but I can't commit the time. Maybe someone is working on
such a thing?

Jens B. Jorgensen

TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
debian-user-request@lists.debian.org . 
Trouble?  e-mail to templin@bucknell.edu .

Reply to: