IETF 75 DTNRG Meeting Minutes [Aug 25; additions possible]
Stockholm, Sweden
31-Jul-2009

IAB Review
	Fall presented review of DTNRG to the IAB on 30-July-2009
	slides are included on meeting tracker

Doc update
	5 RFCs published (4838, 5050, 5325-5327)
	14 "live" (dtnrg) drafts
	4 non expired author drafts
	see DTNRG-4 slide set for per-draft info

Disconnectathon (Ivancic; NASA)
	Interoperability testing took place in terminal room and with other
	participants including NASA, Ohio U, BBN, Trinity, [N4C EU Project]
	implementations: Spindle (BBN), DTN2 (Intel, others), ION (NASA)
	convergence layers: UDP, TCP
	mostly static routing
	some bundles get stuck
	platforms: iPhone, various Linux, Symbian
	used dynamic DNS updates (on Internet) for coordination

Neighbor Discovery (BBN)
	draft-irtf-dtnrg-ipnd-00 (implemented by BBN)
		(UDP/IP specific)
	enumerated or unknown
	possibly across multiple IP hops

Intentional Naming (Prithwish Basu, BBN)
	ref: MIT Intentional Naming (1999)
	think queries in name structure
	progressive resolution
	scalability concerns if aggregation hard
	multiple routing schemes
	see draft-pbasu-dtnrg-naming-00.txt

Network Management (Ivancic)
	draft-ivancic-dtnrg-network-management-reqs-00
		want more comments
	related
		draft-cole-netconf-robust-config-01.txt
		draft-ietf-opsawg-operations-and-management-08.txt
	logging tool

Hermes: DT Bulk Xfers on the Internet (Laoutaris, Telefonica)
	[big] bulk data transfer (e.g. FedEx)
	see http://hermes.tid.es/hermes
	reasons for big data - science data, data mining, dist production, etc..
	goal: use excess capacity (at a good time) for DT bulk data xfers
	expensive networks (like CERN LHC)
		FedEx could be ~20x cheaper
		(today xfer is 27TByte/day)
	approach: water-filling technique over time
		requires in network storage

	discussion as to if FedEx costs will remain over time
		(e.g. as storage density goes up... it is on faster curve)

BBN Discussions of possible Extensions to BP (Leland)
	relative timestamps, hops to live, more CoS bits

  Discussion on Timing

	DTN bundle protocol includes absolute time
	some issues:
		getting the right time
		keeping the right time
		setting the wrong time
	discussion of how hard getting the time really is
		we have watches after all
	resistance by (e.g. some customers) from BBN to set time

	[most time spent here]

   Hops to live

   	motivated by routing loops
	(discussion that bundle ts wasn't designed to avoid this)

	also concern that IP-style TTL gives no info regarding
		hops traveled so far

   CoS bits

	want to map IP ToS bits into DTN CoS but not enough bits

	discussion: but DTN BP isn't designed just to carry IP, so is the need
		really compelling?

Other issues:
	if a WG on DTN subject matter is desired
	having another meeting at IETF 76 seems supported