pan-users
[Top][All Lists]
Advanced

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

Re: [Pan-users] Policy discussion: GNKSA


From: Alan Meyer
Subject: Re: [Pan-users] Policy discussion: GNKSA
Date: Mon, 4 Jul 2011 12:08:36 -0700 (PDT)

Rob <address@hidden> wrote:

> On Monday 04 July 2011 11:42, Alan Meyer wrote:
> > Of course in theory, theory and practice are the same.  But in
> > practice ... maybe I'm totally screwed up here.
>
> Practically speaking, I hit binsearch.info and found an Ubuntu ISO to test
> with, currently the first result here:
>
> http://binsearch.info/?q=ubuntu-10.04.1-desktop-
> i386.iso&max=100&adv_age=600&server=
>
> (you'll need to fix the word-wrapped URL).  I was going to post the result
> of 10 connections, 4 connections and 1 connection, but we have dinner
> guests arriving shortly and the single-connection one is still going.  Each
> time, I created a new directory, put the nzb file into it, purged Pan's
> article cache, adjusted the max connections to the appropriate number and
> watched jnettop to verify it was working after issuing the following
> command line:
>
> time pan --no-gui --nzb ubuntu\ iso.nzb -o `pwd`
>
> I did 10 connections first, so that if there were any server-side caching
> it would be disadvantaged.
>
> real    14m1.951s
> user    1m46.775s
> sys    0m31.602s
>
> Then, 4 connections.
>
> real    35m31.818s
> user    4m37.997s
> sys    1m11.544s
>
> So, you save 60% of the download time using 10 connections vs. 4.  It's
> been 25 minutes and the single-connection one is still going.

Ah well, there's nothing like hard data.  So much for my theory.

Can you tell us (after dinner of course) what sort of computer
you were running on (number and type of CPUs) and the speed of
your Internet connection?

Your results are consistent with server side (or even local
router based) per connection throttling.  If we assume that the
ubuntu binary was around 710 MB, we get the following numbers:

    10 connection throughput:
        710 MB / 842 seconds = 
            843 KB/s total
             84 KB/s/connection

     4 connection throughput:
        710 MB / 2132 seconds =
            333 KB/s total
             83 KB/s/connection

If the single connection version also comes in around 83 KB/s,
then it sure does look like connection throttling, doesn't it?

If so, then we don't know that everyone will benefit from more
than four connections, but it looks like Giganews users will,
unless there's some kind of throttling going on on your side of
the Internet.

If you download a binary via ftp from Ubuntu, or Sun, or some
other site with a known high speed pipe, I presume you'll get
much faster speeds on a single connection.  In that case the
throttling is definitely coming from Giganews, not your router.
That's probably worth a test just to be sure.

What country and state/province are you in?  Anyone else nearby
on this list using Giganews that can see if they are also limited
to 83 KB/s?

It could also be load related.  Gignews might change the values
of their throttling numbers based on the number of connections at
the time.  For example, it might be a lower number on weekends,
or dynamically changing in response to constantly changing
connection numbers.

Of course, all advertising puffery aside, Giganews can't fill
everyone's pipe at unlimited speeds no matter how many users log
on.

    Alan



reply via email to

[Prev in Thread] Current Thread [Next in Thread]