[xcp] Notes from XCP meeting at ISI

[Bob Braden]

19 Feb 2004 XCP Meeting at ISI

Aaron was out of town.


* Two major research areas still open:
	-- Varying link error rates
	-- Effect of mice on XCP  (Need ideas here... Ted tried to dangle it
		as a thesis topic.)

* Checksums
	-- Clarified how TCP checksums currently work with XCP.

	-- Ted's FreeBSD code does not support hardware checksumming.

	-- NOTE: The XCP header is currently not covered by ANY checksum!!
		We should probably fix this!

	-- Ted asked: if XCP header checksum fails, what do you do?
		Bob argued: better drop the packet, to prevent a simple
		attack on XCP.


* XCP header format
	-- Bob asked if it wasn't time to deep-six the cwnd version of the
		XCP header.  Agreed; Eric/Aman will test the throughput
		version, and hopefully be able to easily fix any bugs
		that show up.

	-- The first nibble is now "XCP version #". Bob suggested that
		we should that imagine the header we are designing may
		be more generally useful, for a variety of congestion
		control schemes.  Then the first nibble could be
		registered with IANA as the "Congestion Control Scheme".


* Detecting Non-XCP routers

	-- It will be important to detect non-XCP routers
		in the path.  For this purpose, we need a TTL field in the
		XCP header.  It will be initialized to the IP TTL, and
		each router's XCP will decrement it.  If there are only
		XCP routers, its value at the receiver will match the TTL
		in the IP header.

	-- This does not quite work with MPLS; something special has to
		be done.  MPLS does not decrement the IP TTL, but
		XCP-capable MPLS hops might decrement the XCP TTL.
		Have to teach an XCP-capable MPLS router to not do
		that!

* Kudos and thanks to Eric and Aman!!!