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

Re: failure detection



On Thu, 18 Aug 2005, David Meyer wrote:

	Are you saying that you believe it is scalable for every
	end-system in the Internet (or even every enterprise) to
	have some kind of quasi-real-time access to the
	equivalent of the DFZ RIB somewhere so that they (the
	end-systems/hosts) can make decisions about which locator
	sets to use?

No, I said it's scaleable for enterprise /sites/.

I'm presuming shim6 is to allow a 'split' mode of operation, where end-systems on a 'shimmed' network don't know of shimming, but the shimming is instead on a subset of hosts (at the edges).

I don't know whether that presumption is valid.

	Further, are you saying that you believe it is scalable
	for every end-system in the Internet to *require* such
	information to support site multi-homing?

I didn't and would not say that.

	And BTW, if so,
	how do I communicate this information to my transport
	stack?

Typically, one avails of routing information by not /requiring/ IP protocols to be involved in the details of routing. Ie by layering things correctly.


That's my prime reason for arguing against involving shim6 preemptively in the business of which local locator/source address to use in packets it outputs.

regards,
--
Paul Jakma	paul@clubi.ie	paul@jakma.org	Key ID: 64A2FF6A
Fortune:
It'll be just like Beggars' Canyon back home.
		-- Luke Skywalker