On May 4, 2010, at 11:16 PM, Danny Mayer wrote:
You may be interested in this article written by a couple of chaps in
Australia, just published by the ACM.
http://queue.acm.org/detail.cfm?id=1773943
There's also a PDF of it:
http://portal.acm.org/ft_gateway.cfm?id=1773943&type=pdf
I don't know if any of this will be helpful for tictoc but it's worth
sending out and their RADclock looks interesting.
This is an interesting article, without really much in the way of
details. A few things that strike me
- They seem to be achieving performance at
the level of 10's of microseconds (in the lab) without any router
assistance.
- they focus on "difference clocks" instead of absolute clocks. (In
other words, this is more syntonization than synchronization.)
- they use a "feed-forward" approach, what I would call a paper clock
(where the underlying clock is not adjusted, and time corrections are
applied to its readings) and argue that this provides better stability
in an network environment.
- they don't allow for meshes, and so have no way to bound path
asymmetries
Regards
Marshall
Danny
_______________________________________________
TICTOC mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/tictoc
_______________________________________________
TICTOC mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/tictoc
