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

suggestion: different sequential method

by Guest on 2016/01/24 10:17:58 PM    
It will be possible to make a sequential mode up to 50% from left to right, and others even sequentially in the same torrent from right to left?

I think about it and maybe it can inhibit the cause effect asking only block left right in aggressively!

I liked it, much of this method, although it is not 100% effective because they still calls out of sequential order parts even in aggressive mode, thought of doing 50% to 50% inverted, will be an option for me validates


Maybe your method of release, ultra-fast series reinforced to receive faster yet so who knows?
It is possible KD?

A small question, I am using sequential method, and shipping super seeding mode, work well right?

I was using your Tixati v2.32 alph test..

Thank you in advance for attention
by Bugmagnet on 2016/01/26 04:13:03 AM    
For downloading files from a swarm of other peers, anything sequential is harmful to you and the swarm. Anything and everything sequential.

The perceived advantage of being able to preview some video aside, any method that impairs the protocol design to randomize pieces and give preference to the rarest first will cause your complete download to be delayed and in some cases impossible. That is caused by any number of schemes to bypass the genius built into the bittorrent protocol and/or a tendency not to share equitably. The impatient need for instant gratification, selfishly thinking of your own immediate desires, will mask the fact that such tricks will not help you in the long run.

Example: Someone seeding a 1 GB file will send only 2 GB total, effectively 2 times the file size, and then will stop seeding. If there are 10 peers trying to download that file, as long as everyone is sensible and does not try to game the system by requesting sequential order, each of the 10 peers will have a random and different 20% of the torrent, making 2 full copies available in the swarm after the original seed stops their transfer of it. If everyone is then patient, all will eventually be able to get the full torrent.

If these 10 users make requests for sequential transfers front and back, then all 10 users will have the same 10% of the front and 10% of the back of the torrent. No one will have the middle 80% content and when the seed stops their torrent, no one will be able to finish the torrent.

Right now I have a huge torrent of over 1000 files at 99.995% complete. A single file is missing 50mB and it sits there week after week unfinished. I have seen it often over the years that we have 20+ peers all at 90% or above and none have the missing 10% needed. I have some torrents from 2013 that have not been able to finish because members of the swarm do not have all the pieces. Sequential downloading contributes to situations like that.

It has been said many times over the years. Sequential downloading is BAD. Simple as that. Why is this concept so hard to understand?

Or is it simply even if they do understand the consequences, too many short-sighted people do not care and just want immediate attention to their personal needs, oblivious to the fact that their behavior is detrimental to themselves and others?

http://forum.utorrent.com/topic/85013-sequential-torrent/

http://wiki.vuze.com/w/Sequential_downloading_is_bad

due to the very nature of a Torrent. If it were possible to do so then I would imagine Torrents would lose their viability for a method of downloading files quickly
http://superuser.com/questions/668562/how-to-download-files-in-torrents-sequentially

Sequential downloading defeats one of the primary goals of torrents: to make rare data common. If only one peer has chunk ae986f6ea789 of a torrent, it makes sense that you'd want to download that chunk first in case the peer disconnects before it's fully seeded. Even with movies/music, 70% of the peers may have the first 70% of the movie downloaded, so if you join the swarm it would make sense to throw you at the final 30% in order to increase potential bandwidth to those chunks. I also see overall speed going down as too many peers are requesting the same chunks, when they could be getting other chunks at higher speeds
https://news.ycombinator.com/item?id=7176218

.                        BBEEEUUUUUUAAAAAHHHHH
.
.            %%%%%%
.           %%%% = =
.           %%C    >
.            _)' _( .' ,
.         __/ |_/\   " *. o
.        /` \_\ \/     %`= '_  .
.       /  )   \/|      .^',*. ,
.      /' /-   o/       - " % '_
.     /\_/     <       = , ^ ~ .
.     )_o|----'|          .`  '
. ___// (_  - (\
.///-(    \'   \\ b'ger

(best viewed with fixed font like courier)
by Guest on 2016/01/26 10:26:36 PM    
Have you ever told me this, but this method would work better than aggressive sequential normal because would seek parts of both sides to meet would total 50% and 100% without any break!

Then I explain when only a very limited upload seeder, uploading random and all other users have the same part? Also never seen? You will say that this is bad for examination?

Why then is not so Super seeding default activated? He is no better than upload without it enabled?

I like to download and sequential I enable super seeding the torrent, help to upload rare piece of get also do it in the wrong way?

The many more situations that you do not mention and also cause the curse torrent exam!


So to say that the best method is always the same and connected to the logic always be rare. No improvements in upload or download?

Also find that upload X would be bad for torrent network?

Or even upload a percentage? In order to launch the torrent?

Best User rare piece to the finish, and best ask User to begin, would also be bad?

All this idea are feasible, but not tested because I do not know how to program torrent, but I ask to test the Tixati will be too much to ask?

Test them and see the result in practice better / worse?

Tixati say it has ultra fast algorimos how do you know that? Upload can be broken down as requested piece? Ordered sequential ask other customer declines upload speed for me because the piece required is not rare? User do that?

I know Tixati to client disposal slow upload auto mode shape, what better?

You know when a piece is no longer rare? From that number? I do not know the torrent, but I know that in 5 piece it is no longer rare to have the 5 User ..
by Guest on 2016/09/23 07:23:31 PM    
You can see rarity in the basest sense on the Details tab (or main view) when looking at Availability Available
Anything under 1.0 means there is no seed. If however like above. You have clients with different parts of the transfer... For ex a compilation. It is possible to "revive" a dead swarm by including that missing file, or pieces of a file.
This is my bane for downloading generic myfile.iso instead of the descriptive names usually appended my_file_from_vacation-2016-09-23.iso
Also notice for each seed you connect to, Available increase by 1.0 This is slightly masked from "super seed peers" But you can see those peers which continue to increase in % but don't actually ask anything from you. With 0 Rem B/s from others.
by Guest on 2016/09/25 05:30:17 AM    
There's different levels of awful when it comes to Sequential Downloading.
1.A piece should be sequentially downloaded, especially if it's being downloaded from only 1 peer or seed at a time. This is a good thing, especially with 4 MB piece sizes...even allows the pieces to be written to drive in 1 big write which is great for SSDs and keeps fragmentation on HDDs to a minimum. This means fast lines can run more torrents at once without crippling their drives and/or cpus. It's a shame Tixati has such terrible performance in this regard:
https://forum.tixati.com/support/1833/
"the allocating files thing in tixati always takes much longer than other programs"

https://forum.tixati.com/support/2520/
"Tixati v2.38 downloading ...couldn't sustain 100 KB/sec under seemingly unhindered conditions!"


2.A torrent can be sequentially downloaded with no harm if there's only 1 peer and 1 or more seeds...and the peer-now-seed sticks around to seed for awhile. All pieces were equally rare in that case. Transmission client does that.

3.Sequential Downloading isn't so kind if there's more than 1 peer:


4.Worse still is streaming a video torrent, which requires extremely high download speed for 1080p or 4k high-framerate.
If/when streaming fails, bad things can happen even to the streamer:


A 1080p video torrent (5 GB size) might need ~20 mbit/sec upload speed from other seeds/peers constantly to the streamer alone.
A single ADSL 2+ connection is lucky to have 1 mbit/sec total upload speed and probably cannot devote it to a single peer on a torrent.
With the old BitTorrent average of 4 upload slots per torrent, even assuming these ADSL 2+ peers/seeds are only running 1 torrent each they're only giving 1/4 mbit/sec to any particular peer.
But if there's more than 4 other peers, they won't even be uploading all the time to the same 4 peers.
So under those conditions, for the streamer to reach 20 mbit/sec download speed there needs to be roughly 80 ADSL 2+ peers/seeds at a minimum!
4 streaming peers at once could "drain dry" the combined upload of 100 ADSL 2+ seeds.

5.There is an even much worse case scenario than this...
A BT client that makes streaming default behavior and has trouble uploading to others that becomes very popular...
...and allows sequential downloading/streaming on multiple torrents at the same time.
by Guest on 2016/09/25 08:26:47 AM    
Since the other-server links were removed, I'll fill in what's missing here...
"3.Sequential Downloading isn't so kind if there's more than 1 peer:"

Sequential downloading:
1.Hammers seeds and peers for "common" pieces, especially the 1st and last pieces in a torrent...when everyone else in the torrent swarm may be desperate for "rare" pieces near the end of the torrent. Seeds will be forced to upload the same starting pieces in a torrent to each new sequential downloader that arrives, leaving less time+upload for these seeds to send rare pieces to other peers.
2.Penalizes other peers trying to play fair (those peers prioritize downloading rarest pieces first) -- because sequential downloaders will usually not be interested in rare pieces and will not upload as much to other peers playing fair *because they gave it little*!
3.Do not even "play nice" with each other! 2 "pure" sequential downloaders (that only download the next piece in sequence) will NEVER Tit-For-Tat (give something to get something, a core component of the BitTorrent protocol) with each other since at no time will both peers have a piece the other wants. So the one that gets further ahead will be pure leeched by ALL the other sequential downloaders as they try to catch up. That means the one in the lead will have to download even more from seeds that should be giving out only rarest pieces!
4.Seeds will be unable to use super seeding/initial seeding mode because sequential downloaders may refuse to download or will be very slow about sharing their rarest piece with other sequential downloaders.
5.Sequential downloading can slow down a torrent swarm to the point that it dies early if/when the last seeds leave after reaching their upload ratio. The remaining peers on the torrent will be very likely to not have enough pieces between them to complete the torrent, even if there's more than 10 of them. This becomes an extremely bad problem on torrents over 1 GB in size. I've seen the last few pieces in a torrent become exceedingly rare even when only about 2 out of 10 peers were sequential downloaders.


"4.Worse still is streaming a video torrent"

Being able to play a video torrent as it downloads requires a download speed faster than the playback speed (unless the download has a long head-start) and MUCH higher than most seeds and peers can upload, so it's unsustainable if everyone does it.
Stopping the torrent very shortly after playback finishes is likely, so the last few pieces of the torrent won't have enough time to be uploaded to others. The already-overloaded (or missing) seeds will be the only ones with the last few pieces in the torrent...so other peers will take much longer to completely download the torrent.

This only works ok where the torrent swarm is overwhelmingly seeds with very few peers...such as on private trackers.
But streaming can fail without warning on many public torrents, even ones that have more seeds than peers...especially if the streamer is firewalled.
Streaming fails when playback catches up to a piece you haven't downloaded yet even if you have a lot of pieces after the missing one.

"If/when streaming fails, bad things can happen to the streamer:"

Depending on the playback method, a missing piece could result in either a blank screen for a few seconds/minutes (if the file was initially filled with 0's), a portion of another video file previously stored in the same spot (which can be confusing or embarrassing), lots of garbage on the screen, a player crash, or file system/OS crash. Worst-case I can think of (but not likely) -- important accessing files get corrupted during a file system/OS crash...rendering the computer incapable of rebooting until the OS is reinstalled.
by Guest on 2016/09/26 02:19:57 AM    
I have a self-coded bot that makes tixati auto-set all transfers to sequential, as I like watching things the moment it starts. In payment I seed on seedboxes for years afterwards to keep the swarms happy. I will never distribute it as I know many others would be much more selfish.
by Guest on 2016/09/26 11:01:33 PM    
Or using any kind of macro keystroke program yes? Or is the bot you're talking about utilizing the WebUI




This web site is powered by Super Simple Server