[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Trigtran BOF

My sense is that the critical resources are not in place for this >>work to succeed, although the work itself may be worthwhile.
Well, as Mark said, there was a good show of hands for people
interested in the "link up" problem.  I think that the working
group probably has the expertise and the resources to work on the
"link up" problem, starting with a thorough problem statement
about the need, the implications of different link layers,
an evaluation of the "implicit communication" approach,
tradeoffs between explicit and implicit communication, etc.
Let me be a bit more specific on what I think is missing. Clearly two new chairs are needed, at least one of which should have had previous experience as a Transport WG chair (insert the usual suspects here).
That should take care of the required transport expertise.

However, I also think that they need to come up with a security advisor. The problem of securing the notifications they described is a difficult one, and it's not clear that it has an obvious solution. I can easily see a group without a security advisor floundering on that alone. In particular, I don't think that IPsec is a viable choice in a situation like this. If the router is one hop away, then perhaps the work going on in SEND could help, but I think they want something more general than that.

Finally, there are some tricky L2 media issues I'm not sure they understand, and they need to bring someone in (like Phil Karn) who can contribute there. Link up and down notifications need to be dampened, and in my experience the appropriate damping interval is of the order of the RTO (or larger!). For example, while RTOmin is typically 1 second, 802.11 trigger damping constants are of the order of 10 seconds, which implies that several retransmissions would need to occur before the type of solutions they are talking about could take effect. I'm concerned that they may not take the time to fully understand the L2 media they are dealing with, such as the nature of the triggers and the damping required for them. This could lead to a "solution" that is worse than the problem itself.

But I agree that it might be one of the more iffy working groups.
It is "iffy" not because of the subject matter, but because of the proposed approach. With the right people and advisors involved, I could see this work succeeding.

_________________________________________________________________
STOP MORE SPAM with the new MSN 8 and get 2 months FREE* http://join.msn.com/?page=features/junkmail