Bram Cohen Lambastes Avalanche

[news=http://filesharingtalk.com/vb3/logos/bt.gif]Late last week, there was a flurry of media attention surrounding Avalanche. Avalanche was labeled as Microsoft's answer to BitTorrent, prompting such headlines as "Microsoft launches Avalanche, rivaling BitTorrent" and "Microsoft Develops BitTorrent Clone." While the mainstream media heralded this as the second coming of Jesus, it turns out that no such application even exists.

"First of all, I'd like to clarify that Avalanche is vaporware," Bram Cohen writes in his weblog. "It isn't a product which you can use or test with, it's a bunch of proposed algorithms. There isn't even a fleshed out network protocol. The 'experiments' they've done are simulations."

On the surface, Avalanche appeared promising. One of the more frustrating aspects of BitTorrent is that many times, file transfers slow down to a crawl towards the end. This happens when the sources of information begin to dry up. Avalanche proposed a type of "Par file" error correcting system. During BitTorrent transfers, a file is broken up into many segments. Avalanche proposed that each segment of a file will be encoded with an error correcting code. Once enough of these codes combined, the file will automatically finish creating itself. Seems like a novel idea, right?

However, Bram Cohen had already covered the topic in November of 2004, in which he responds to repeated requests for the feature.

"It isn't done because, quite simply, it wouldn't help."

Bram goes more into technical detail by presenting three possible benefits, with corresponding rebuttals.

Bram goes on to criticize Avalanches implementation of error correcting.

"The really big unfixable problem with error correction is that peers can't verify data with a secure hash before they pass it on to other peers. As a result, it's quite straightforward for a malicious peer to poison an entire swarm just by uploading a little bit of data. The Avalanche paper conveniently doesn't mention that problem."

Bram continued to rip into the very heart of Avalanche, dismantling most of their proposals. He did however, credit the work on from a more literary point of view.

"As you've probably figured out by now, I think that paper is complete garbage. Unfortunately it's actually one of the better academic papers on BitTorrent, because it makes some attempt, however feeble, to do an apples to apples comparison."

Bram Cohen's journal.


http://filesharingtalk.com/vb3/image...ews_source.gif Source: Slyck[/news]

Quote:

---

The really big unfixable problem with error correction is that peers can't verify data with a secure hash before they pass it on to other peers. As a result, it's quite straightforward for a malicious peer to poison an entire swarm just by uploading a little bit of data. The Avalanche paper conveniently doesn't mention that problem.

---

typical m$ :lol:

I figured there might be a problem in that area when I saw the original article, but I didn't know if there was enough info to get round that problem, possibly by transmitting the hash of combined blocks that you know to be correct. Of course who can say that you aren't one of the ones poisoning the swarm, so that probably doesn't work.

An equally (possibly even more) serious problem is the amount of computing/disk power/memory to combine/split the blocks before sending/after receiving.