Brad Beyenhof wrote:
> I'm not sure what you mean by PFM, but I'll do my best to answer your
> questions. Firstly, the differing speeds in different clients is
> probably because you're doing the download at different times. The
> landscape of P2P availability is constantly changing as people connect
> and disconnect to/from the web of sharers. However, different clients
> may have different defautl settings (discussed below).
>
>   
PFM = "Pure Fscking Magic"

And as for download at different times, well, transmission was sitting
idle with 2 peers connected, I switched to the same torrent on azureus
and immediately got about 25K from 20 peers. Why the difference?
> It could be a client configuration issue that only conencts to a
> certain number of sharers per torrent (or total across all torrents).
> However, it's also possible that downloading from less than the total
> available number of peers saturates your available downstream
> bandwidth, or that the tracker is making sure to only give you the
> connections that are most reliable and/or closest in proximity to your
> current location.
>
> In addition, where did you see the number of seeders listed? If it was
> on the Web site where you found the .torrent file, those results are
> usually pretty stale.
>
>   
There is a config for download speed and upload speed, that's all, I max
out the download to unlimited, the upload I cap at 64K. Peer info can be
found in the details portion  of most clients Azureus actually seems to
give quite a bit of info on the transfer. I figure it's getting this
info from the tracker and so should be up to date.

> Again, the speed issue is largely a function of how much of the file
> is available. It's possible to have great speeds when you start a
> torrent since everybody connected has some portion of it, but many
> torrents have certain pieces of the file present in a limited number
> of places (usually caused by people closing the torrent before they've
> seeded it enough). In fact, I've run into several torrents where only
> a portion of the file is actually available despite several peers:
> people will full copies of the file (or, the only ones with the
> missing pieces) had dropped out.
>
>   
All parts of the file show as available, and even were they not, it
should not affect the download speed of the available parts.
> P2P is great for bandwith and distribution costs, but for the
> end-users the file is often only fully available following its initial
> release. Too many selfish downloaders, and the system degrades to the
> point of unusability. Thus, the creation of authenticated trackers
> (like tvtorrents.com) that keep track of your total seeding ratio and
> make sure you stay on the positive side if you want to keep using the
> service.
>
> In any case, Linux distribution torrents are usually pretty well
> seeded, as the open-source ethos lends itself to being more generous
> with one upstream bandwidth. That, and people like me who keep
> torrents seeding on their servers for a while after the download is
> finished... I dare say that non-Linux folk are much less likely to
> have always-on server machines.
>
>   

And I get all that, I just want to understand why my downloads seem to
be so slow when none of that applies. With lots of seeders and even
leech peers, I should be getting much better speeds, or at least I think so.

Tyrion


-- 
KPLUG-List@kernel-panic.org
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list

Reply via email to