utorrent pro leecher???

[Xaero]

What is this I am hearing now (or reading), first you let utorrent quit as soon as the downloads are complete and now the download limit is gone!

This is bad news. Considering the sheer size of utorrent/bitorrent users I am sure you will see a leecher uprising in the torrent network.

These's features should be removed. This is bad for the torrent network. Quit when everything completes is ok but not when downloads complete.

[thelittlefire]

The Queueing rules have been in since...

The removal of the multiplier was a bad step IMO also.

[Xaero]

Removing of the download cap may not make much of a difference because from what I have seen if you cap your uploads you anyway will get less download speed. But that's not the whole point!

Everything that makes people think "no upload, no download" should remain. It is the only way to keep leeches from breeding in the torrent network. How does removing this make utorrent better? Maybe for a few it might be helpful but not for the millions of others.

[Switeck]

It's part of a bigger problem. Even out of EVERYONE who posts on this message forum...of those that come here wanting help with their speed problems...very few mean their UPLOAD speed problems. Almost all want faster download speeds and at most make only a token claim to care about their upload.

In short, they'd sabotage their upload speed if they thought it would speed up THEIR download speed. And many think just that! Why else would they want to set upload speed to 1 KB/sec for new torrents?

When the global download limiter came out for v1.4, there was a major outcry for a short while...but it was easy to bypass.

The per-torrent download limiter which lasted all of v1.8 only wasn't so easy to bypass, so the complaints were considerably worse. Those that complained the most often said it was private trackers' job to make people share.

But the BitTorrent protocol itself says it's mandatory to share. It may say much about download/upload limits, but it does say limit connections + upload slots to allow good TCP flow control AND that the purpose of BitTorrent is so ALL participants can complete the torrent download...not just those who download OR upload the quickest. Not everyone is likely to complete torrents if hit-and-run downloaders can set their upload speed to 1 KB/sec, download quickly, and leave often without sharing 1/100th as much as they downloaded. :mad:

So if this limit is gone for good, uTorrent's/BitTorrent's tit-for-tat system needs to be rethought and strengthened. This may mean uTorrent needs to use a download/upload strategy more like the "selfish" BitTyrant client...that will basically only upload (quickly at least) to those that upload back to it. BitTyrant's goal was to spend its limited upload speed wisely...so it would maximize how much download speed it got back. It would ALWAYS try to upload at full speed...but more in a trader fashion than a giving/sharing fashion. Once in seed mode, it would upload much like any other BitTorrent client -- to almost anyone.

The optimistic upload slot has its place, but once it has given a particular peer a whole piece to share...it really shouldn't keep giving more to that peer UNTIL it has given to most other peers as well. Repeatedly giving to a peer that does not return the favor just reinforces bad peer behavior, poor quality swarms, and bad torrent survivability!

[Xaero]

Uh........???

So you are saying it is OK it remove the caps or not?

[Switeck]

Uh...I wasn't clear enough?

"Not everyone is likely to complete torrents if hit-and-run downloaders can set their upload speed to 1 KB/sec, download quickly, and leave often without sharing 1/100th as much as they downloaded."

and

"So if this limit is gone for good, uTorrent's/BitTorrent's tit-for-tat system needs to be rethought and strengthened."

Removing the caps was not well thought out. I was in total shock and denial when I heard it was being removed.

I almost feel it was a knee-jerk reaction to all the hostility heaped on by those who didn't care for it. Also, it was viewed as a "draconian" measure that people might just abandon uTorrent over.

Given an "honor system", people are generally dishonorable...because it's easier. Given an incentive system, people will do whatever benefits them the most...so long as its easy. The caps were an easy incentive.

The damage repair necessary to make up for the lack of caps cannot be done by private trackers alone.

[Ultima]

So if this limit is gone for good, uTorrent's/BitTorrent's tit-for-tat system needs to be rethought and strengthened.

Firon is pushing for a better choking system to be implemented. What constitutes as "better" remains to be seen

[Xaero]

Ok, but untill the better choking system is implemented?

[Ultima]

To clarify, the operative keyword in my previous post was "pushing," meaning that there haven't been any official plans made to do so (just Firon talking about it). The point is that changes to the choking mechanism wouldn't be completely out of the question. As such, if/when it'll be implemented isn't known.

[Switeck]

Currently, the BitTorrent protocol is "pull" based. Peers pull stuff from seeds. The peers decide what they want to download from the seed, and the seed can do little more than send it (slow or fast) or not set it. This means 10 peers in a row can request the same start+end pieces of the torrent, and once done can only download more off the seed since they all have the same parts. Beggars shouldn't be such picky choosers.

A seed with many peer connections should be just as good or better than peers at judging what the rarest pieces of its portion of the torrent swarm are...than peers which have likely not been connected as long or to as many peers. Until seeds have more control over which pieces get uploaded next, (such as how initial seed mode works) leeching-without-uploading will be VERY damaging to torrent swarms.

...Because once a seed leaves, all the rare pieces instantly get a lot rarer! When the last seed leaves, the torrent may become unfinishable for the remaining peers.

Implementing BitTorrent FAST Extensions should help slightly to reduce leeching indirectly. Because much like Initial Seeding, Seeds and Peers will "work harder" to keep the torrent's availability high by uploading predominately rare pieces.

[Xaero]

I noticed something strange with utorrent today: when I am browsing I usually limit the download speeds so that I am able to download without any problems and usually if I am download a file I limit the download speed to 5Ks on utorrent. The upload is set to unlimited.

But after a while the upload speeds also seem to settle around 5ks. If I cap the downloads to 10k then the upload would avg around 10ks.

I don't understand the logic behind this. Usually everyone gains uploads why does utorrent cap it? It doens't really show a limit but I can clearly see from the graph that the upload and download speeds around the same level.

This brings me to another interesting question: does this happen vice versa as well? I'm sure it does because I have noticed on numerous occasions that increasing uploads would increase downloads and that's a good thing. But the other way round is bad.

So what happens if you have a 2MBps/256kpbs asymmetric connection? And what about all the talk of having upload speeds at 80% of max. All this doesn't fall into place. Any explanation for this strange behaviour?

[Switeck]

That's an unfortunate side-effect of the peculiarities of TCP/IP networking. Capping upload speed limits not only how fast you're uploading but also how fast you can send out requests to seeds and peers to KEEP uploading to you. They don't get those requests, they assume they or you are overloaded...and slow down.

[Xaero]

And capping download speeds does the same as well?

So what about a 2MBps/256kpbs asymmetric connection?

[thelittlefire]

Download speeds peter out in an inverse based upon how many connections you allow. uTorrent tries to equalize the download (limit) across all peers to not overtax any one of them.. this in turn I guess lowers the algorithm's decision to UPLOAD to a peer even though you're only LETTING it send you data so fast.

[Xaero]

Would that mean in case of a 2MBps/256kpbs connection?

[thelittlefire]

I don't know. What are you trying to limit your download to out of a theoretical max of 256, likely sustained 220-235? 256 Kbit means theoretical 32, and sustained usually @ 26-28.

[Xaero]

What I am asking is: if someone has a broadband connection with max download of 2mbits/s and upload of 256kbits/s i.e., they will not be able to upload faster than 32kpbs but can download at 256kpbs would their download be limited around 32kbps?

[thelittlefire]

Not usually. Only with very bad settings do you see a degradation in download speed over time. With rotating peers in a large swarm you should be able to hit your sustained speed (ISP issues notwithstanding).

[Xaero]

How is that possible? Aren't you supposed to cap your upload speeds to 80% of max to allow protocol overhead? That would mean that you would have to set the upload limit to around 26kbps. Wouldn't that limit the download rates as well?

[Firon]

Too high of an upload = saturated connection: slower downloads.

Too low of an upload = other peers don't wanna give you bandwidth: slower downloads.

You have to find the right balance.

[Xaero]

I noticed something very very very strange in utorrent...

Try this..

When downloading a torrent...

1. cap the upload to 1kb/s and unlimited download and check the download speeds

2. cap the download to 1kb/s and unlimited upload and check the uploads speeds!!!

3. now enable to scheduler in seed only mode and check the upload speeds

I just can't get my head around this. It's very interesting can someone explain why this happens? I'm using utorrent 1.8.1.

[Switeck]

TCP flow control cannot work properly if either download or upload is artificially limited to very low speeds.

[Firon]

Download capping, unfortunately, also seems to restrict protocol communication. In effect, a low cap will choke all communication and prevent proper uploading.

[Xaero]

What I am saying is that it will upload at the same speed at the download cap is set. The reason I asked to set a cap of 1k is so that the difference is clearly visible in the speed graph. You don't really have to set a low cap 50% will suffice. You will see the same result.

[Switeck]

I set my download speed max to 1 KB/sec in uTorrent v1.8.1 and upload speed remained unchanged.

I was actually quite surprised.

However I am only seeding torrents, not downloading any at the moment.

And I don't have estimate overheads (net.calc_overhead) turned on.