headers
Aniruddha Bhalekar | 3 Jan 2003 00:14
Picon

[e2e] Pipelining in HTTP connections


I am under the impression that most of the popular browsers (with an
exception of Opera) donot support HTTP pipelining of requests. I heard
otherwise about Netscape v7.0 and IE v6.0 but on reading the release
documentation, didnot find anything that supported that.

Can anyone shed any light on this? Thanks a lot!

regards,
Aniruddha.

-------------------------------------------------------------------------------

				ANIRUDDHA BHALEKAR
			   Graduate Research Assistant
		Center for Satellite and Hybrid Communication Networks
			       University of Maryland

aniruddha <at> ieee.org	      www.isr.umd.edu/~anibha	    	(301) 919 0867

-------------------------------------------------------------------------------
Permalink | Reply | 2 comments |
headers
Dan Wing | 3 Jan 2003 01:31
Picon
Favicon

RE: [e2e] Pipelining in HTTP connections

Mozilla supports pipelining,
http://www.mozilla.org/projects/netlib/http/pipelining-faq.html, and I have
it enabled on my Mozilla (via its configuration GUI - no mucking around with
configuration files).  So far I haven't found a site that doesn't work with
pipelining.

I don't know if Mozilla is on your list of popular browsers.

I expect Netscape 7 also supports it, but they may have removed the GUI
setting for it.

-d

> -----Original Message-----
> From: end2end-interest-admin <at> postel.org
> [mailto:end2end-interest-admin <at> postel.org]On Behalf Of Aniruddha
> Bhalekar
> Sent: Thursday, January 02, 2003 3:14 PM
> To: end2end-interest <at> postel.org
> Subject: [e2e] Pipelining in HTTP connections
>
>
>
> I am under the impression that most of the popular browsers (with an
> exception of Opera) donot support HTTP pipelining of requests. I heard
> otherwise about Netscape v7.0 and IE v6.0 but on reading the release
> documentation, didnot find anything that supported that.
>
> Can anyone shed any light on this? Thanks a lot!
>
(Continue reading)

Permalink | Reply | 0 comments |
headers
David P. Reed | 3 Jan 2003 02:12
Picon

RE: [e2e] Pipelining in HTTP connections

I may be incorrect, but I believe that Microsoft IE supports it, calling it 
HTTP1.1 support.

At 04:31 PM 1/2/2003 -0800, Dan Wing wrote:
>Mozilla supports pipelining,
>http://www.mozilla.org/projects/netlib/http/pipelining-faq.html, and I have
>it enabled on my Mozilla (via its configuration GUI - no mucking around with
>configuration files).  So far I haven't found a site that doesn't work with
>pipelining.
>
>I don't know if Mozilla is on your list of popular browsers.
>
>I expect Netscape 7 also supports it, but they may have removed the GUI
>setting for it.
>
>-d
>
> > -----Original Message-----
> > From: end2end-interest-admin <at> postel.org
> > [mailto:end2end-interest-admin <at> postel.org]On Behalf Of Aniruddha
> > Bhalekar
> > Sent: Thursday, January 02, 2003 3:14 PM
> > To: end2end-interest <at> postel.org
> > Subject: [e2e] Pipelining in HTTP connections
> >
> >
> >
> > I am under the impression that most of the popular browsers (with an
> > exception of Opera) donot support HTTP pipelining of requests. I heard
> > otherwise about Netscape v7.0 and IE v6.0 but on reading the release
(Continue reading)

Permalink | Reply | 0 comments |
headers
George Michaelson | 3 Jan 2003 02:43
Favicon

Re: [e2e] Pipelining in HTTP connections


We turned off pipelining. A side effect of keep-alive and related methods for
webservers doing micro-apps is that the overall 'dead child' collection you
have to hold in your webserver rises dramatically, for all the people who come
once, never again, but leave a browser connection live beyond the life of
their micro-service (we do whois on web, which is basically
request-> response-> DIE)

If we could run a non-forked threaded server, it would probably make sense to
turn it back on. Alas, we now also depend on mod_perl, which mitigates against
that (too much saved state in the perl, memory consumption issues, lack of
clue).

it may actually be an overall lack of clue on my part, but turning it off made
the difference between life and death on the webserver. It made a huge and
immediate difference to the number of backlogged child processes waiting to
die off.

So while its maybe nicer end-to-end for long-lived bindings at 'session' level
to the web client/server, its not always a good fit for the server side, if
your jobmix differs.

cheers

-George
Permalink | Reply | 0 comments |
headers
Spencer Dawkins | 3 Jan 2003 15:56

RE: [e2e] Pipelining in HTTP connections

Hi, David,

This is not what we've seen in IE 6.0 - in the IE GUI, "HTTP/1.1" refers to the use of persistent connections,
not just the use of HTTP/1.1 headers, and requests on "HTTP/1.1" connections are NOT pipelined.

No matter how much we wish they were.

In the Netscape 7.0 I've been using, persistent connections and pipelining are two separate settings
(from "HTTP/1.1"), and pipelining has the desired effect, but is turned off by default.

Spencer

> -----Original Message-----
> From: David P. Reed [mailto:dpreed <at> reed.com]
> Sent: Thursday, January 02, 2003 7:13 PM
> To: Dan Wing; Aniruddha Bhalekar; end2end-interest <at> postel.org
> Subject: RE: [e2e] Pipelining in HTTP connections
> 
> 
> I may be incorrect, but I believe that Microsoft IE supports 
> it, calling it 
> HTTP1.1 support.
> 
> At 04:31 PM 1/2/2003 -0800, Dan Wing wrote:
> >Mozilla supports pipelining,
> >http://www.mozilla.org/projects/netlib/http/pipelining-faq.ht
> ml, and I have
> >it enabled on my Mozilla (via its configuration GUI - no 
> mucking around with
> >configuration files).  So far I haven't found a site that
(Continue reading)

Permalink | Reply | 0 comments |
headers
Alaa Youssef | 2 Jan 2003 04:09
Picon
Favicon

[e2e] Infocom 2003 early registration


            *************************************************
            *               IEEE INFOCOM 2003               *
            *   The Conference on Computer Communications   *
            *                                               *
            *    March 30 - April 3, 2003, San Francisco    *
            *                                               *
            *    The 22nd Annual Joint Conference of the    *
            *   IEEE Computer and Communications Societies  *
            *                                               *
            *       http://www.ieee-infocom.org/2003        *
            *************************************************

EARLY REGISTRATION for Infocom 2003 is now available online at
http://www.ieee-infocom.org/2003/registration.htm
Early *discounted* registration ends on February 28, 2003.

Hotel reservations are also available online at
http://www.ieee-infocom.org/2003/hotel_reservation.htm

CONFERENCE SCHEDULE
        Tutorials                        March 30-31, 2003
        Conference                       April 1-3, 2003

For general information please check the conference web site or 
contact the General Chair, Fred Bauer (fredbauer <at> ieee.org).
Permalink | Reply | 0 comments |
headers
Jim.Gettys | 5 Jan 2003 01:51
Picon
Favicon

Re: [e2e] Pipelining in HTTP connections


> From: "David P. Reed" <dpreed <at> reed.com>
> Date: Thu, 02 Jan 2003 20:12:48 -0500
> To: "Dan Wing" <dwing <at> cisco.com>, "Aniruddha Bhalekar" <anibha <at> glue.umd.edu>,
>        <end2end-interest <at> postel.org>
> Subject: RE: [e2e] Pipelining in HTTP connections
> -----
> I may be incorrect, but I believe that Microsoft IE supports it, calling it
> HTTP1.1 support.
>

Nope.  You can be HTTP/1.1 compliant and not support pipelining.

Current Mozilla supports it; see edit->preferences->advanced->http_networking.
It was added shortly before 1.0, and may not be really done well.

Opera also (besides mozilla) supports pipelining; Hakon Lie is their CTO,
and worked with us on our HTTP/1.1 stuff, and I expect does it really
well.  It seemed pretty blazing fast to me.

Last I knew, IE did not support pipelining, but I mostly stay away
from Windoze these days.

Sometime soon (once my busted ankle has healed, which means in a couple
months, unfortunately) I intend to look at some packet traces and see if
moz has implemented pipelining well.  We found in our work that you have
to be very careful to get full performance: with many fewer socket buffers
(and therefore the aggregate of offered windows being much smaller), best
performance is achieved with a single TCP connection that one *very*
carefully aggressively reads from (or you get lots of delayed acks, etc.).
(Continue reading)

Permalink | Reply | 1 comment |
headers
David P. Reed | 6 Jan 2003 16:40
Picon

[e2e] RFI: Microsoft accused of TCP standards violation

The following was shared with me and is making the rounds on 
slashdot.   Before I do the work to verify/refute this, has anyone from e2e 
any input?   I'm personally slightly skeptical that this is correct (though 
of course I'd be really concerned if it were).
------------------------------------------------------------------------------------
Interesting... Microsoft is using TCP/IP in a way that
makes IE respond faster to IIS and slower to non-IIS
webservers:
 > http://grotto11.com/blog/?+1039831658
Could someone with IE and IIS please verify this? A dump of
TCP/IP
traffic with different versions of IE and IIS would be
ideal.
Of course covered on Slashdot:
 > http://slashdot.org/article.pl?sid=03/01/05/2025254
This post sums things up:
The parent +5 post is flat out wrong. This is not about
persistant connections, which is a high-level HTTP feature
that keeps a connection open so that the browser can send
more requests. This is about a low-level TCP hack that IE
uses to get a small speed boost on IIS servers, while
breaking TCP standards compliance.
If I read the article correctly, instead of creating a new
TCP connection and then sending a request, IE sends the
request immediately without bothering to finish the TCP
handshake. Microsoft IIS web servers deal with it
automatically, and it is faster because it saves a
round-trip wait for the ACK and the following requset.
The down side is that non-IIS servers have no clue what this
incoming packet is. It must be invalid because it is not a
(Continue reading)

Permalink | Reply | 0 comments |
headers
Vernon Schryver | 6 Jan 2003 18:08
Favicon

Re: [e2e] RFI: Microsoft accused of TCP standards violation

> From: "David P. Reed" <dpreed <at> reed.com>

> ...
> It could also be a potential security risk, because if this
> is true, then it makes it very easy to IP-spoof a HTTP
> request against IIS (since the request is a self-contained
> packet instead of a long connection sequence).

If it's true, then it becomes interesting to ask about the initial
congestion window used by IIS.  For example, if you send a TCP segment
advertising a 60K window and containing an HTTP request for a large
HTTP document, will IIS blast 40 packets at a victim forged as the IP
source address?  Or will it send only 2 or 4 segments?

It sounds easy to test, and so wrong-headed that it must all be false.
Even the embrace and extenders in Redmond are that foolish...well,
not always.  Some of their PPP "enhancements" weren't far off.

Vernon Schryver    vjs <at> rhyolite.com
Permalink | Reply | 0 comments |
headers
Marc Slemko | 6 Jan 2003 19:32

Re: [e2e] RFI: Microsoft accused of TCP standards violation

On Mon, 6 Jan 2003, David P. Reed wrote:

> The following was shared with me and is making the rounds on
> slashdot.   Before I do the work to verify/refute this, has anyone from e2e
> any input?   I'm personally slightly skeptical that this is correct (though
> of course I'd be really concerned if it were).

Wow, what a huge amount of technical garbage in that whole /. thread.
What else is new.

This speculation is all based off:

	http://grotto11.com/blog/slash.html?+1039831658

I think someone just got confused because they didn't understand
TCP very well, and didn't understand that IIS defaults to keeping
persistent connections open indefinitely (ie. no timeout) if there
are ample connections available on the server, and didn't understand
that even if you think you have closed IE, you really haven't.  So
sure, a persistent connection will stick around when some people don't
expect it.

The optimization that IIS does to not stick an arbitrary hard
timeout on persistent connections isn't a bad one as far as I can
figure.  (the ugly thing is that they provide no way to force a
limit, and don't even provide any option to disable persistent
connections, but that is another story)

Now, it may be that there is a bug in IE and/or some versions of
windows that prevents it in some situations from properly noticing
(Continue reading)

Permalink | Reply | 0 comments |

Gmane