Transport protocols and bulk file transfer

[Larry Dunn <>]

Scott,
Interesting, but I'd "wait and see" - look at an on-network implementation.

Feeding in network-assumptions to get a set of rules to try,
and acknowledging that if you include the assumption that
"buffer filling may occur" drives the algorithm to be comparable to
existing schemes, seems to suggest there are lots of real
scenarios where the improvements will not look
as great as the graphs suggest.
Also disconcerting that the authors note that they
don't understand how/why it seems to "outperform" XCP (which
has it's own issues, long-ago documented in conversations w/ Dina...).

I'll read the full paper when I've got time.

Thanks for the pointer- great to see work in the area!

Larry
--


On Jul 20, 2013, at 2:57 PM, Scott Brim wrote:

> http://web.mit.edu/remy/
> 
> thoughts?
>