Help and Support
Ask a question, report a problem, request a feature...
<<  Back To Forum

Feature Concept Suggest to Maximize Bandwidth

by overnightplumbus on 2017/06/07 04:23:17 PM    
Ok... So each torrent has a set amount of allowed peers.

What I am thinking is add a feature to for example take the amount of Bytes In and compare against the Time Connected to clear peers from your connected Peer pool that are below a certain rate of data transferred compared to how much time has elapsed.

This will remove peers that are transferring on the low end of the scale and would be best implemented with number ratio or something similarly to Seed ratio settings. This purge of connections can happen once every 10 minutes for example to allow connections to stabilize after being established...

I was manually sorting through my 75 connected peers and adding them to the ignore list periodically and managed to get a download going from around ~200k to up to ~1800k consistently using this method but it had to be achieved manually.

I think rolling this out in a future TIXATI update as a new option would be great for Torrents that have a large peer pool but are running greatly reduced in speed. Out of the 75 peers I had, to start I had about 4 peers that were each feeding me about 50k with the majority of peers only running around 4kb/sec each. Through using this method with the ignore functionality in the connected peer tab I was able to multiply the speed I was pulling ten-fold.

Saved me from doing a 3 day download down to 12 hours for a new torrent yesterday, thought it would be good to bounce the idea off the team.

-overnightplumbus
by Bugmagnet on 2017/06/08 04:24:13 PM    
so what you are saying is that poor people, with limited internet speeds are not welcome?
by Guest on 2017/06/08 06:25:19 PM    
Not one mention of poster's UPLOAD speed at the time... so we don't know if the upload-to-download exchange rate was really bad (and representing a bug that needs to be fixed since it means Tixati's Tit-For-Tat BitTorrent implementation is broken) or the poster is just a leech wanting to hit-and-run quicker. Or both.

When your download is slow, think about the other side of the equation!
How do others see your peer or seed on a torrent?
How fast are you uploading? How many upload slots (peers) are you uploading to at once?
If you're uploading less than 5 KB/sec per upload slot, you typically WON'T get a good download speed in return. (Tit-For-Tat tries to reward only the big givers.)
So download speed can increase by finding a balance -- reducing upload slots if too many and/or raising upload speed if too slow.

Once a seed, are you connecting to LOTS of peers but uploading to very few of them?
Simple solution -- reduce max connections per torrent once only seeding and/or increase upload slots.
by Guest on 2017/06/08 08:40:20 PM    
overnightplumbus, get on private trackers instead of suggesting ideas that go against the very concept of the bittorrent protocol.
by ZarkBit on 2017/06/08 11:53:38 PM    
Not such a bad idea, I constantly struggle with the same issue, limited max peer and those who are connected are seeding at very low speeds, that really could maximize the chance of getting a good peer.

On the other hand, that would negatively impact users on private trackers, since a large majority has low upload speeds, takes some time for them to hit a good ratio, they would be discarded with this feature and start to have difficulties hitting that ratio (talking as if Tixati was greatly used among private trackers :P).

As for non private trackers, not sure how this could have a negative impact, since ratio isn't a thing.

Tixati unfortunately already has a bad reputation among private trackers, and this would only hurt the client even more, could be considered a discriminatory feature against people who have low upload speeds.

But I'd really like other's peoples opinions on this one, seems interesting.
by BRMateus2 on 2017/06/09 02:30:30 AM    
Doesn't the mainline clients already has disconnect when not the best bandwidth peer? If I'm correctly about that, the client already has an "(change peer, close connection and open new conn.) when not good";
by overnightplumbus on 2017/06/09 07:40:34 AM    
I've got a non assymetrical line that is 18/3 so my upload speed is always shiiiiteee. Would be a nice feature for those that aren't lucky enough to have a nice upstream!
by overnightplumbus on 2017/06/09 03:07:37 PM    
I have used this method on a couple torrent that are well over 5GB in file size. With my experimentation I ranged from 20 - 200 connected peers while running a single download on my 18/3 connection that has been costing me over $100 monthly lately being that after my first 1024 GB I am billed $10/5gb. So basically I'm being charged as much as it would probably cost me in US Dollars as it would to have Fiber internet and monthly expenses paid for like Ukraine (rough estimation from random people commenting around reddit here or there)...

So anyway more about the testing, back on topic. And as far contributing and being a leech is concerned I have been involved in file sharing scene for a long time and I used to scan for public FTP sites and test them for bandwidth, available space, and if the site operator would retain or purge the files I would use to initially test. From scanning, to filling, to organize and maintaining an up to date selection of files, I did it all, when hardly anyone had cable internet and you were lucky to have a 56k connection and get over 15kb/sec with a dial up connection. I remember downloading a preview of Windows 98 in like 96 and being extremely excited that my 600MB file downloaded in a mere 2 days lol... So I have done my fair of sharing accordingly to the *hot* file sharing method at the time. So don't call me a leech or say I am circumventing the torrent protocol when I suggest this method to increase speed and recommend this being added to TIXATI before another client does it. This could be some edge pushing add on and if it really hurts your feelings to get the most out of your download connection, I am sorry but CHANCES ARE you probably aren't the one suffering with the poor asymmetrical connection... This is just a good method to choose faster peers.

So now more on topic about the topic.... In reply to an earlier reply... This wouldn't be used to abuse private tracker connection, as they should be limiting you to your contribution speed, this would be for a high peer rate torrent, say >1000 peers, >50 seeds, on a public tracker....

I found my optimum peer count for running just 1 torrent at a time, with a focus on getting that download as fast as I possibly can for reasons directly relating to IMPATIENCE!!, to be between 20-60 peers.

* Slowly climbing to a steady average ofa round 200kbps I gave the peers about 10 minutes to stabilize.

* I sort the Peer list in the peers tab by "B/s In" taking care to note any peers that may not currently be high up on current speed but have a large "Bytes In' field. (This was critical because my fastest peer of all, that came through numerous purges of my peer fishing bank, would occasionally drop to speeds that I was purging, and this type peer was critical was critical in nearly every test. For one torrent this sporadically changing peer accounted for over 12GB out a 36.074 GB file).

* Every 5 - 10 minutes I would select, right click and ignore those peers until I could find a connection I was happy with out of each individual peer.

* Each time with about 10 purges total per test, I was able to get a 200-400k download to a speed of 1200k-1800k per second. Additional tests were run with the same torrent running dry (ignored peers re-enabled), same peer count as my active ignore based tests (60), and also 200 peers which is the max in TIXATI. For the most part these hardly reached 1/4th the speed as my active ignore, peer selective pruning method.


SO THE FACTS ARE THERE. WORKS 100 PERCENT OF THE TIME 90 PERCENT OF THE TIME! jkjkjkjkjkjk

!!! No but for realsies... This is a good way to get that torrent that you are itching for just that much faster. And although some may frown upon this method, in reality I am still sharing the file just as much when it comes to B/s and total Bytes Out !!!

!!!!! I'm not saying to eliminate all upload speed together and only select peers that are seeds, YOU ARE STILL SHARING THE SAME AMOUNT OF DATA IN THE END SURELY !!!!!


So developers, programmers, etc. etc. etc....

IMPLEMENT THIS! Someone else eventually will! Or maybe allow add on scripts? (Does anyone know of a torrent client that would support programming an add on or script that would do the same thing within the software?)

This would be the bees knees guys.. I mean come on. If you make a torrent client and you're not rolling out changes to make your downloads faster than the other downloads, your gonna have a bad time. Thank you for your time in reading my ramblings and remember, if you pizza when your supposed to french fry, you're gonna have a bad time.
by BRMateus2 on 2017/06/10 04:01:17 PM    
I support that, even if I don't have those problems with an 20/2 Mbps (2072KBs/256KBs averaged) connection.
Though that was already implemented in an soft way.
As for damage, this should be soft, not aggressive.
by Guest on 2017/06/10 05:59:59 PM    
This mentality widespread would have real damage on torrent network health as a whole. If everyone did it, watch torrents become as slow as Tor given time.
by Guest on 2017/10/01 01:58:05 AM    
Tixati needs better choke/unchoke logic.
It doesn't need to "cheat"...it's just junk there.

It could even borrow BitTyrant's logic for exploiting available connected seeds and peers.
There's a few university-level papers on the subject...
by Guest on 2017/10/16 05:50:13 AM    
This might be totally obvious, but the real thing that P2P software should be focused on is to maximize total upload speed.

So what I know about BitTyrant (not much though) it has some merits and possibly that same strategy could be used if slightly modified.

While I like to download fast, uploading fast is much more important to me. And uploading effectively so that the material spreads as efficiently as possible. Thats because I seed stuff that I think is very very important, and because Im seeding 10TB constantly, so while I have large upstream, it still is very little per share.

Tixati has been the fastest uploader so far if measured in raw upstream bit flow. Thats the reason why I use it. Any upgrade to uploading logic, that makes the total swarm uploads faster, is greatly appreciated.

More and more people are more focused on the seeding each day. Those people would probably donate to support the development of better uploading strategy in Tixati. Whether it would be "upload maximization" option or embedded inside completely.
by Shellsunder on 2017/10/24 04:55:54 PM    
by ZarkBit on 2017/06/08 05:53:38 PM
Tixati unfortunately already has a bad reputation among private trackers
What leads you to say that judgment is still true?  I haven't heard anything negative about Tixati in a long while.


OP
by overnightplumbus on 2017/06/09 09:07:37 AM
What you did in the distant past hardly matters in this.  You say you were at one time very concerned with others' wellfare and with making info distribution more effective, that's commendable, but now, your strategy belies that.

The criticisms and comments of these posters are true and strong.
Bugmagnet on 2017/06/08 10:24:13 AM
Guest on 2017/06/08 12:25:19 PM
Guest on 2017/06/08 02:40:20 PM
ZarkBit on 2017/06/08 05:53:38 PM
Guest on 2017/06/10 11:59:59 AM

It's clear that if Tixati would aggressively eliminate peers who contribute less than you'd like, the whole cooperation in P2P would suffer and only those with high bandwidth contributions would get much use out of it.


by Guest on 2017/10/15 11:50:13 PM
I seed stuff that I think is very very important
That seems such a broad or presumptuous claim that it should be justified.  What are some examples of the material you seed that is so very important?
by ZarkBit on 2017/10/25 09:43:59 PM    
What leads you to say that judgment is still true? I haven't heard anything negative about Tixati in a long while.
Experience.




This web site is powered by Super Simple Server