Olaf van der Spek
olafvdspek at gmail.com
Sat Apr 2 12:25:30 EST 2005
On Apr 1, 2005 6:09 PM, Simon Spero <ses at unc.edu> wrote:
> Any proposal for using a custom transport protocol for bulk data
> transfer instead of TCP must meet a very high standard of proof.
> Modern implementations of TCP represent the accumulation of decades of
> experience, and it has met and absorbed many challenges over the years.
> A look at the history of TCP Vegas and XTP would be informative.
> At the very least any such transport *must* be tested under ns2 and be
> shown to be fair with respect to TCP, and to not trigger any congestion
What or who 'requires' this?
More information about the BitTorrent