[Mod_gzip] http 1.1 or 1.0 which to accept

mod_gzip@lists.over.net mod_gzip@lists.over.net
Fri, 19 Mar 2004 13:53:41 +0100


Dies ist eine mehrteilige Nachricht im MIME-Format.
--=_alternative 0046D9BCC1256E5C_=
Content-Type: text/plain; charset="US-ASCII"

Hi bob,


> Right now we only allow 1.1 http for gzip compression.
> We turned on compression but don't see much of a drop
> in bandwidth.

mod_gzip isn't something that you can just "turn on" or "turn off".
You need to understand what it will do for you and which of its
many effects will be beneficiary and which ones will not (or even
counter productive, as the "Vary:" issue might turn out to be in
certain situations).

Up to which detail do you understand what mod_gzip is actually
doing for you? You might want to have a look at
        http://www.schroepl.net/projekte/mgzta/mgzta.html
as to understand what I'm talking about. If you had some report
like this then you would understand the effect of each one of
your configuration rules, as well as the characteristics of your
visitors.
(During the years about five people ever asked me for this
script so I never took the time to make a documention or even
a download package - yet thinking this way is what I would
suggest every mod_gzip user to do for his/her own installation.)

> A lot of access to our site is people at work and I am
> wondering if most corporation firewalls, proxy server
> and the like are changing http to 1.0.

I am using mod_gzip in a scenario where one proxy server is
enforcing the use of HTTP/1.0 for the whole server cluster.
Yet, this has nothing to do with compression efficiency.

And in the WWW I have experienced that more than 50% of
my visitors are using M$IE with HTTP/1.0 and thus not even
sending "Accept-Encoding: gzip" so that any server effort is
just wasted time. I cannot force anyone to understand his/her
browser configuration.

Then again, on some commercial servers I am using mod_gzip
for I have some mechanism similar to
        http://www.schroepl.net/projekte/msie/
running that will be embedded into certain pages. Therefore
some customers at least are aware that they could do better
but it is still their choice to do anything or leave it as it is.
And talking to our customers made them consider changing
their browser configuration in certain cases so that now 85%
of all requests are sending "Accept-Encoding: gzip", which
allows for saving about 50% of the total traffic (including
HTTP headers!).

Regards, Michael
--=_alternative 0046D9BCC1256E5C_=
Content-Type: text/html; charset="US-ASCII"


<br><font size=2 color=red face="sans-serif"><b>Hi bob,</b></font>
<br>
<br>
<br><font size=2 color=red face="sans-serif"><b>&gt; </b></font><font size=2><tt>Right
now we only allow 1.1 http for gzip compression.</tt></font>
<br><font size=2 color=red face="sans-serif"><b>&gt;</b></font><font size=2><tt>
We turned on</tt></font><font size=2 color=red face="sans-serif"><b> </b></font><font size=2><tt>compression
but</tt></font><font size=2 color=red face="sans-serif"><b> </b></font><font size=2><tt>don't
see much of a drop</tt></font>
<br><font size=2 color=red face="sans-serif"><b>&gt;</b></font><font size=2><tt>
in bandwidth.</tt></font>
<br>
<br><font size=2 color=red face="sans-serif"><b>mod_gzip isn't something
that you can just &quot;turn on&quot; or &quot;turn off&quot;.</b></font>
<br><font size=2 color=red face="sans-serif"><b>You need to understand
what it will do for you and which of its</b></font>
<br><font size=2 color=red face="sans-serif"><b>many effects will be beneficiary
and which ones will not (or even</b></font>
<br><font size=2 color=red face="sans-serif"><b>counter productive, as
the &quot;Vary:&quot; issue might turn out to be in</b></font>
<br><font size=2 color=red face="sans-serif"><b>certain situations).</b></font>
<br>
<br><font size=2 color=red face="sans-serif"><b>Up to which detail do you
understand what mod_gzip is actually</b></font>
<br><font size=2 color=red face="sans-serif"><b>doing for you? You might
want to have a look at</b></font>
<br><font size=2 color=red face="sans-serif"><b>&nbsp; &nbsp; &nbsp; &nbsp;
http://www.schroepl.net/projekte/mgzta/mgzta.html</b></font>
<br><font size=2 color=red face="sans-serif"><b>as to understand what I'm
talking about. If you had some report</b></font>
<br><font size=2 color=red face="sans-serif"><b>like this then you would
understand the effect of each one of</b></font>
<br><font size=2 color=red face="sans-serif"><b>your configuration rules,
as well as the characteristics of your</b></font>
<br><font size=2 color=red face="sans-serif"><b>visitors.</b></font>
<br><font size=2 color=red face="sans-serif"><b>(During the years about
five people ever asked me for this</b></font>
<br><font size=2 color=red face="sans-serif"><b>script so I never took
the time to make a documention or even</b></font>
<br><font size=2 color=red face="sans-serif"><b>a download package - yet
thinking this way is what I would</b></font>
<br><font size=2 color=red face="sans-serif"><b>suggest every mod_gzip
user to do for his/her own installation.)</b></font>
<br>
<br><font size=2 color=red face="sans-serif"><b>&gt; </b></font><font size=2><tt>A
lot of access to our site is people</tt></font><font size=2 color=red face="sans-serif"><b>
</b></font><font size=2><tt>at work</tt></font><font size=2 color=red face="sans-serif"><b>
</b></font><font size=2><tt>and I am</tt></font>
<br><font size=2 color=red face="sans-serif"><b>&gt;</b></font><font size=2><tt>
wondering if most corporation firewalls, proxy server</tt></font>
<br><font size=2 color=red face="sans-serif"><b>&gt;</b></font><font size=2><tt>
and the like</tt></font><font size=2 color=red face="sans-serif"><b> </b></font><font size=2><tt>are</tt></font><font size=2 color=red face="sans-serif"><b>
</b></font><font size=2><tt>changing http to 1.0.</tt></font>
<br>
<br><font size=2 color=red face="sans-serif"><b>I am using mod_gzip in
a scenario where one proxy server is</b></font>
<br><font size=2 color=red face="sans-serif"><b>enforcing the use of HTTP/1.0
for the whole server cluster.</b></font>
<br><font size=2 color=red face="sans-serif"><b>Yet, this has nothing to
do with compression efficiency.</b></font>
<br>
<br><font size=2 color=red face="sans-serif"><b>And in the WWW I have experienced
that more than 50% of</b></font>
<br><font size=2 color=red face="sans-serif"><b>my visitors are using M$IE
with HTTP/1.0 and thus not even</b></font>
<br><font size=2 color=red face="sans-serif"><b>sending &quot;Accept-Encoding:
gzip&quot; so that any server effort is</b></font>
<br><font size=2 color=red face="sans-serif"><b>just wasted time. I cannot
force anyone to understand his/her</b></font>
<br><font size=2 color=red face="sans-serif"><b>browser configuration.</b></font>
<br>
<br><font size=2 color=red face="sans-serif"><b>Then again, on some commercial
servers I am using mod_gzip</b></font>
<br><font size=2 color=red face="sans-serif"><b>for I have some mechanism
similar to</b></font>
<br><font size=2 color=red face="sans-serif"><b>&nbsp; &nbsp; &nbsp; &nbsp;
http://www.schroepl.net/projekte/msie/</b></font>
<br><font size=2 color=red face="sans-serif"><b>running that will be embedded
into certain pages. Therefore</b></font>
<br><font size=2 color=red face="sans-serif"><b>some customers at least
are aware that they could do better</b></font>
<br><font size=2 color=red face="sans-serif"><b>but it is still their choice
to do anything or leave it as it is.</b></font>
<br><font size=2 color=red face="sans-serif"><b>And talking to our customers
made them consider changing</b></font>
<br><font size=2 color=red face="sans-serif"><b>their browser configuration
in certain cases so that now 85%</b></font>
<br><font size=2 color=red face="sans-serif"><b>of all requests are sending
&quot;Accept-Encoding: gzip&quot;, which</b></font>
<br><font size=2 color=red face="sans-serif"><b>allows for saving about
50% of the total traffic (including</b></font>
<br><font size=2 color=red face="sans-serif"><b>HTTP headers!).</b></font>
<br>
<br><font size=2 color=red face="sans-serif"><b>Regards, Michael</b></font>
--=_alternative 0046D9BCC1256E5C_=--