ietf
[Top] [All Lists]

RE: BitTorrent (Was: Re: [Isms] ISMS charter broken- onus should beon WG to fix it)

2005-09-16 06:43:49
Wouldn't it be good if an ISP could install a machine that would
function as a local head end BitTorrent cache?

There are probably multiple folk on my comcast drop who are viewing the
same feeds from CNN and Crooks and Liars etc. Peer to Peer is not the
optimal way to support this but it is the optimal way to bootstrap.

This is of course leading to a discussion where someone says 'you want
to work out the optimal routing path, that's hard' and someone else says
'I thought we were meant to be the ones with brains the size of a
planet'.

Don't bother with optimal, just do SRV lookups on the reverse DNS. The
network operator can do the optimal cache placement and use the reverse
DNS to advertise the correct location. 

-----Original Message-----
From: ietf-bounces(_at_)ietf(_dot_)org 
[mailto:ietf-bounces(_at_)ietf(_dot_)org] On 
Behalf Of Thomas Narten
Sent: Friday, September 16, 2005 9:24 AM
To: Scott W Brim
Cc: Michael Thomas; Paul Hoffman; ietf(_at_)ietf(_dot_)org
Subject: Re: BitTorrent (Was: Re: [Isms] ISMS charter broken- 
onus should beon WG to fix it) 


Scott W Brim <sbrim(_at_)cisco(_dot_)com> writes:

The metaphor I'm trying to use this week is that the IETF is 
landscapers and we provide a fertile, beautiful area for 
people to go 
wild and create excellent gardens.

Exactly. The beauty of TCP/IP (and indeed many protocols when done
well) is that they are generic enablers for additional 
higher-level uses.

TCP/IP creates opportunity for innovation, and does so in a 
way that is generally safe for the network.

In the case of BitTorrent, it runs on top of TCP. It is silly 
to assume/expect all application protocols to be developed in 
the IETF.

It is true that BitTorrent (or more precisely its heavy use) 
creates interesting dynamics that have implications for the 
net and maybe even the IETF.

For example, BitTorrent creates an environment in which end 
users start running "background" jobs that run for hours and 
suck up idle background network capacity. I've heard ISPs use 
figures of 30% or more of their capacity.... This is Just 
Fine at one level, but also upsets some business models.  
Wouldn't it be nice if BitTorrent traffic (at least in some 
cases) could be labeled as "background" traffic so that ISPs 
had the ability to better prioritize or figure out when it is 
critical to add more bandwidth vs. "just nice to have"? Maybe 
more work here for diffserv?

Thomas

_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/ietf



_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/ietf

<Prev in Thread] Current Thread [Next in Thread>
  • RE: BitTorrent (Was: Re: [Isms] ISMS charter broken- onus should beon WG to fix it), Hallam-Baker, Phillip <=