Maximum number of seeds on torrent

[krydda01]

I have just moved from using Vuze/Azureus to uTorrent.

But there's one feature in Vuze which I can't find in uTorrent:

On each torrent, you can specify the maximum number of seed connections.

See: http://wiki.vuze.com/w/UG_Advanced_Information#Options_tab

It's very useful if you want to improve your ratio while downloading. If you have maximum 50 connections, often you don't need more than half of them as seeds anyway.

I have looked in the FAQ and in the forum, but haven't seen this. Sorry if I missed it, but please let me know where to find it if that's the case.

Thank you

krydda01

[Ultima]

It's very useful if you want to improve your ratio while downloading.

I really wonder how much this feature would actually help in this regard. Your assumption here is that other peers would already have pieces that you downloaded from the seeds, in which case you wouldn't be able to upload what you downloaded... but that probably isn't the case for a large chunk of the time.

Looking at it from a different perspective, if you don't get rare pieces from the seeds, then maybe other peers wouldn't download from you either, as they may already have every other piece you have. In that case, this option would be detrimental to your ratio, since you'd have to download all new pieces from other peers only, and not be able upload back to them either.

Too, upload slots are limited, so really, you wouldn't be able to upload to most of the peers at any given time anyway.

Overall, I only see such a feature as having the potential to introduce artificial efficiency bottlenecks into swarms, since it closes yet another path through which rare pieces could potentially be obtained.

[krydda01]

Ultima wrote:

Your assumption here is that other peers would already have pieces that you downloaded from the seeds, in which case you wouldn't be able to upload what you downloaded

Actually, my assumption is the opposite; that a number of the peers (leeches) don't have the pieces that I have downloaded, which they then can get from me.

My knowledge on bittorrent, peers, seeds and leeches is a bit limited, so I don't follow your reasoning completly, but changing this setting in Vuze does the trick for me.

When I don't have this parameter set, having a total of 50 peers, I have close to 50 seeds and close to 0 leeches. When setting the maximum seeds to 20, the number of seeds goes down and leeches goes up (obviously). The downloading speed isn't really effected, but the uploading clearly increases.

Of course, I could use Vuze when I need this feature. I feel uTorrent is superior in all other aspects which are interesting to me, so it would be good for me if this could be introduced.

[Ultima]

Actually, my assumption is the opposite; that a number of the peers (leeches) don't have the pieces that I have downloaded, which they then can get from me.

Your assumption here is that other peers would already have pieces that you downloaded from the seeds ... [emphasis added]

If in general you assumed that you had pieces that other peers didn't have, including any you downloaded from the seed, then this option really wouldn't be all that helpful, as you'd only be cutting out the latter -- pieces you could download from the seed that no one else has, and which (in turn) you'd be able to upload to other peers (were you connected to the seed).

At any rate, I can't say anything conclusively, as this isn't something that's within my domain of expertise (if I have any). It's really down to statistics and game theory (operations research), which would probably mean extensive tests/experiments. If it worked for you, then I can't claim it doesn't work at all. My main concern is how this would affect swarms in general if it became more widespread in use.

[krydda01]

Ultima wrote:

My main concern is how this would affect swarms in general if it became more widespread in use.

I understand. The last thing I'd want to do is mess up the swarms.

It's really down to statistics and game theory (operations research), which would probably mean extensive tests/experiments.

Ok, now I'm starting to feel silly. My intention was not to add a heavy work load on you guys.

All in all, it's maybe better to remove this thread. I'd hate to have given people any ideas that potentially will cause seeding problems.

Thank you for taking the time to respond!

[Ultima]

Oh, the request is still perfectly valid. To implement it doesn't require all of that research; all of that research would only qualm my personal fears (which may be completely ungrounded).