On 21-Jan-2007, at 07:14, Alexander Harrowell wrote:
Regarding your first point, it's really surprising that existing
P2P applications don't include topology awareness. After all, the
underlying TCP already has mechanisms to perceive the relative
nearness of a network entity - counting hops or round-trip latency.
Imagine a BT-like client that searches for available torrents, and
records the round-trip time to each host it contacts. These it
places in a lookup table and picks the fastest responders to
initiate the data transfer. Those are likely to be the closest, if
not in distance then topologically, and the ones with the most
bandwidth. Further, imagine that it caches the search - so when
you next seek a file, it checks for it first on the hosts nearest
to it in its "routing table", stepping down progressively if it's
not there. It's a form of local-pref.
Remember though that the dynamics of the system need to assume that
individual clients will be selfish, and even though it might be in
the interests of the network as a whole to choose local peers, if you
can get faster *throughput* (not round-trip response) from a remote
peer, it's a necessary assumption that the peer will do so.
Protocols need to be designed such that a client is rewarded in
faster downloads for uploading in a fashion that best benefits the
swarm.
The third step is for content producers to directly add their torrents
to the ISP peers before releasing the torrent directly to the public.
This gets "official" content pre-positioned for efficient
distribution,
making it perform better (from a user's perspective) than pirated
content.
If there was a big fast server in every ISP with a monstrous pile of
disk which retrieved torrents automatically from a selection of
popular RSS feeds, which kept seeding torrents for as long as there
was interest and/or disk, and which had some rate shaping installed
on the host such that traffic that wasn't on-net (e.g. to/from
customers) or free (e.g. to/from peers) was rate-crippled, how far
would that go to emulating this behaviour with existing live
torrents? Speaking from a technical perspective only, and ignoring
the legal minefield.
If anybody has tried this, I'd be interested to hear whether on-net
clients actually take advantage of the local monster seed, or whether
they persist in pulling data from elsewhere.
Joe
