[BLKMGK]

That's not how bittorrent works at all...

Actually yeah that pretty much IS how it works. Anyone in your torrent cloud sees the IP addresses of (most) everyone else. Seeds and downloaders alike. Next thing you know you've got people joining the streams just to get the IP of everyone else. Any public tracker ala TPB is rife with this!

Torrents also work best on large files, many small files, which these images would be, waste bandwidth with overhead. The overhead might be as much as the files themselves I'm afraid.

Some sort of P2P way to distribute data is a great way to remove burden on community sites i agree but I'm not sure that this is the right way...

How about hosting the files on usenet? It would require access to the service and takedowns, as usual, might screw with it but that could solve the issue of sharing IPs...?

[Ned Scott]

Torrents wouldn't show "the exact contents of your video library to the entire world", only what is specifically being shared for the torrent. Also, in this case it wouldn't even tell anyone if you have the whole movie or not. XBMC allows media stubs (wiki) to be scraped, and various other things can use scraper data that don't actually lead to watching the movie itself (pull metadata for trailers, listings, etc).

[pecinko]

Locally storing images for library items, not doing multiply fanarts and posters stuff might help.

If/when XBMC server arrives it will help further as you will to DL images on every client but rather just once on server.

[tvdb-scott]

Odd that this thread is gaining traction again. I assume it's because we've had some speed issues lately. As of right now, and for the foreseeable future, TheTVDB is NOT shutting down. Back when we posted this we received a decent amount of donations that stabilized us a bit. I'm not saying that we'll never need to look at additional revenue streams like premium features or API fees for large commercial projects, but rather that we've buffered away from a complete shutdown.

To people suggesting that we simply move to a $100/month host with unlimited bandwidth, I'm sorry but it doesn't work that way. The very large majority of dedicated providers allow unlimited bandwidth ONLY until you start degrading the performance of the other sites they host. How do I know this? Because we've gone that route in the past and been shut down very quickly. Based on our discussions with different providers we'd be looking at somewhere between $1000-1500/month for a server and connection capable of handling our site.

Right now we're using Cloudflare, which is very nice for offloading bandwidth. To give you an idea of the scale we're dealing with, over the last 30 days we've had 3.2 billion requests for 88TB. That amount is growing fast, and will likely grow even faster once Raspberry Pi based consumer devices start being released. We're still dealing with a VERY high load on our MySQL server and are discussing solutions.

[nickr]

Thank you for the update szsori. Good project, keep up the good work.

[DukeOfTheWicked]

Kickstarter has proven itself to raise LOADS of money for projects. And to my knowledge, it doesnt have to be a new project trying to take off. Why not find a new server that is efficient and cheap enough, calculate what it would cost to fund 5 years worth of hosting and ask for that amount... Then put that project on Kickstarter. It'll be project "save thetvdb"! Sounds reasonable to me.

[Buzniak]

If the TVDB gets shutdown, it will be a huge dent in XBMC. That's the only scraper i use for TV shows. BEST ONE FOR SURE.

[maliqua]

let people host mirrors
problem instantly solves itself.

[nickr]

let people host mirrors
problem instantly solves itself.

How? Who will host these mirrors? Who will pay for the bandwidth?

[maliqua]

let people host mirrors
problem instantly solves itself.

How? Who will host these mirrors? Who will pay for the bandwidth?

lots of people would musicbrainz does it. no one mirror would ever see the kind of bandwidth the main site sees so it would be trivial if db dumps or something similar where available
88tb a month is not really that much bandwidth which is the trivial part to be honest the db load is likely more of an issue. so even if the people hosting mirrors were taking large bursts of bandwidth to fetch the mirror even if it was disproportional to the amount of use it would offload in raw bandwidth the compute and io offload from less queries could still be substantial.

[nickr]

We look forward to your distributed tvdb database implementation, anyway no point debating such proposals here, go talk to the tvdb owners

[mrevilbreakfast]

This is just a high-level thing but...

• TVDB server database is expanded to include space for version of current data for the record of each TV show and also a separate table that holds the unique record identifier for each TV show and the IP addresses of the PCs that host copies of that.
  
• User's XBMC/TVDB client makes request for data relating to program X.
  
• TVDB database responds, advising a list of a number of randomly chosen IP addresses of similar clients that host this information.
  
• TVDB client polls those IP addresses until one replies with the data, and transfers it.
  
• On completion TVDB client verifies file integrity and size, and if deemed successful, updates TVDB server database so that that TV show's DB entry records its IP as having a copy.
  

Clients refresh their connection every X hours/days so that the TVDB database is up-to-date with live clients.
Data served therefore goes from being the size of the graphics and text to the size of the metadata served to it.
In theory, if people leave their media PCs on for 8 hours a day, and there's a spread across the globe the TVDB will only ever have to serve the full TV show data once.
TVDB therefore only has to serve "new" data plus any data that no longer has vitality, ie, there's no live copies of it.
Change control can be maintained by applying version data to newly served data, so when there's an update on the TVDB servers clients can pull down the new version, allowing for replication.

Conscious that that's a high level description, but it could work that way!

[Death-Axe]

I would donate if there wasn't so many arrogant assholes running the site. (that hikore guy or whatever his name is needs the boot). It is a great site for scraping and everything, but really as a community driven site, they need to stop being assholes to the users.

[tential]

I would donate if there wasn't so many arrogant assholes running the site. (that hikore guy or whatever his name is needs the boot). It is a great site for scraping and everything, but really as a community driven site, they need to stop being assholes to the users.

Explain? I've never used their forums or had to update anything so I don't know.

[crash123]

As long as you follow the rules and be courteous there are no problems. If you don't, we let you know.