ISPs should embrace and cache bittorrent packets, not throttle it

badcheese.com

10 points·scumola··6 comments

The article suggests that HTTP caching is "a fairly common practice nowadays." The claim is totally unsupported by any data and is likely incorrect. Most of HN probably have access to a web server on the Internet. Try throwing up some static content and refresh the page a few times. Notice any missing hits in the access logs? I doubt it.

There would be no need for CDNs if it were cheaper for ISPs to build a massive caching proxy infrastructure than it is to just buy more bandwidth and routers. I don't see how that statement is any less true for bittorrent than it is for HTTP.

Keep in mind that P2P consumes upstream bandwidth which is precious in many ISPs. If caches can reduce that, the economics may change.

See related discussion in IETF ALTO WG: http://www.ietf.org/mail-archive/web/alto/current/maillist.h...

I'm not able to follow that discussion very closely (lots of technical ISP jargon in there that I don't understand), but I guess the concern is the ISP caching copyrighted content, right? Isn't the caching of copyrighted images in an http proxy just the same issue though?

Isn't the caching of copyrighted images in an http proxy just the same issue though?

That's the point that I made, and Weaver's response is (1) there's a difference between what's legal and what will get you sued and (2) HTTP caching is in some sense grandfathered in, while there is still time to define rules for P2P caching.

It would have been nice to see an informed discussion of the likely hit rate for such a cache as a function of its size, taking into account both keys and data.

Reply on news.ycombinator.com