Minutes IETF110: dtn
minutes-110-dtn-00

IETF DTN working group meeting
IETF 110 - Virtual

Thurs 2021-03-11 17:00 - 19:00 (UTC)

Co-chairs: Rick Taylor, Ed Birrane

Notetakes: Adam Wiethuechter

Agenda
=====
- Admin, Chairs, 10 mins.

Start at 1100
cRT: Initial welcome
cEB on the agenda, 2 parts
    Part 1, 4 presentations on deployment
    Part 2, next steps for rechartering and prioritization of items
cRT: please be cautious with notes!
    Take care to avoid colliding with others :-)
    Magnus leaving, welcome Zaheduzzaman! (Zahed)
    Sec is Adam W!
    Lots of docs in RFCEditor queue and on their way to standard!
        bis, sec and tcp-cl4
    BPSec needs AD Review
    Charter item for Custody
        Superseeded by BIBE
    RT will glady present if needed

- Bundle-in-Bundle Encapsulation (BIBE), 15 mins
  (https://datatracker.ietf.org/doc/draft-ietf-dtn-bibect/)
  Scott Burleigh

Talk from 3 years ago as no changes since then and good summary
Original BIBE from 2009 from MITRE Corp.
Capability of BP nodes agent (as application)
    Support Content-centric networking
    Still valid motiivations
Resurrected in 2013 as a convergence layer
    Intends to disentagle routing from security
    BIBE takes place of security source/dest
Custody Transfer in 2015/16
    Removed from BP ultimately
    Could not be made effecient (estimate rtt not possible)
    delay tolerant asym needed in some scenarios
Custody transfer at convergence layer node to node
    BIBE did this already!
    Cross-domain security and reliable coverage
Aggregate custody signing
    2012 group from U of Colorado, Boulder
    Used for ISS for asymm data rates
ACS success, replaced old custody transfer in BIBE
    May 2018 draft posted with this change
Reliable convergence layer protocol
    Details on how it works
Applications
    Will not go through all of them here in detail
Future
    BIBE consider adoption and part of charter
    Spec simple and not hard to implement
    Powerful

cRT: Noted that want WG adoption, roll into charter discussion

- BPv7 tests with ESA OPS-SAT, 15 mins
  Marius Feldmann and Felix Walter

Title changed
MF:
    Background
    Ring Road concept
        For low cost comm networks
        Two fights; Dresden and ESA OPS-SAT
        Issues!
            CubeSat needed simple impl. - wouldn't make it due to complexity
            ESA OPS-SAT no feedback?
        Time moved on and microPCN grew and became open source
    microD3TN
        was first BPv7 impl.
        modular design
        App Agent Protocol - simple integration
    Field Test 1: Project COLDSUN
        uD3TN used as first field test of BPv7
        02/2019-07/2019
        Mobile underwater network through sat.
        Answer to not having flight opportunity!
    ESA OPS-SAT
        Dec 2019, D3TN first experiment
        Experiement to show ring road
        uD3TN both on ground and sat.
        Planned extended version with ION
FW:
    Shed some light on tech details for experiement with ESA
    Had to config all services running
    Example of a D3TN configuation
    Shows some log files of an actual test
    Conclusion
        BPv7 works great!
        Routing and forwarding via timestamps are error prone
        Standardized config/management required
            Would like to explore AMP in future
        Tools to check neighbor availability in DTN massively useful!
    More tests and experiements together please!

cRT:
    Very interesting and conclusions will feed into discussions in second half

- Potential updates to UDPCL definition of RFC 7122, 15 mins
  Brian Sipos

Updated to UDPCL
    Speculation and open to question
    UDP provides unique behavior that other transports don't have
    Main goal: backwards compatability 1 bundle = 1 datagram
        BPv6 and v7 compatability?
    Weak points of UDPCL
        BPSec and certificate chain the bundle would approach MTU
        Unicast vs Multicast
        BPAgent level details
            Must agent choose interface bundle be transported on?
    Security
        Here is how, but don't need to
        Send and forget strategy
    Bundle packing
        in datagram from effeciency
        rn 1 bundle = 1 datagram
        what about packing multiple if useful?
        agent behind NAT, udp convo could advertise
    Draft posted
        DTLS use and other topics
        github link...
    Next Steps
        need to discussion above things
        wg draft?

MW:
    Assume maintain simple? Maybe want to discuss reuse of ...?
BS:
    Did look into these topics, consistent port nums, etc. UDPCL does not
    mention these things.
MW:
    bidirectional? multicast?
BS:
    Nothing about return path in current draft. Is there interest running
    behing NAT
RT:
    With magnus, take stuff from NAT punch tech. DTN has uses for messaging
    services and will run into NATS and things.
MW:
   No paticular interest or need, but was asking and suggests keeping it
   simple. Recommend against.
cRT:
    Sort of topics that WG would sort out, it is interesting work.

COSE Contexts
    Similar as last presentation
    Glue between two existing protocols and well defined
    Proposes a simple set of profiles
        Sym and Asym keys
    Asym key behavior in little effort as possible
    COSE nice as no bi-directionality
    COSE WG with PKIX and X509
    Main requirements already exists
    Demo implementation with bi-directional

cRT: WG adoption as well

- Interplanetary Internet interoperability and management considerations, 15
mins
  Alberto Montilla

Pilot project WG
provide IPNSIG their view and what they are doing
Currently building DTN testbed with various active participants
Would like to keep IETF WG up to date with progress and learnings (this is
first) Active projects
    DTN testbed
        networks of networks for experiementation
        looking at how we scale, 100-1000s of nodes
        interoperability
        new feature testing
        lots of member projects
            network management for space and earth
            extending Crisis information mangement systems (earth)
                using drones
            AI-based image recongnition
DTN Testbed
    By Feb 4 orgs onboard
    Welcoming other implementations for interop
    Exchanging messages, files
    Started multicast
    Management is critical
        Hard to manage and error
        One EID that was wrong every test
        Not just in networks but between
        Seperate routing from management
Network Management
    Very early draft of ... that orgs are using
    Transfering contacts automatically
Example of Unified Medical Records for Space Exploration
Reindeer husbandry
    Fully operational
    Propreitary version of DTN
Commitment to Standards
    BPv7
    AMP for management framework
Summary
    DTN testbed in progress
    Learning a lot
        management and routing options
    Wish to continue reporting

cRT: please continue reporting!
    Scale like this is great feedback

- Next Steps for the WG / Re-chartering, 40 mins
  Chairs

cEB:
    In prep we looked back at new agenda topics for past few years
    105 and 106 in prep, calls were put out for next steps
        Lots of topics came in
    Concerns
        This is a lot, how do we trim down?
        Already stuff that was active
            BIBE, UDPCL, Key Admin, Asyn Admin, COSE Sec
    Process
        Walk through list
        Anything missing?
        Priority for next charter
        Topic by topic here, then to mailing list
        111 we make final decision (unless need more time)
        Interims might be needed
    Topic breakdown
        Extensions to existing work
            More and closer to being finished
        New work on top of BP and DTN Arch
            Spend time scoping and understanding on fewer of these
    Extensions
        BIBE, Manfiest Block, Convergence Layers, Security Contexts, Async
        Management
    Async Management
        Vital to manage devices in DTN network
        cRT: this has been brought up 2-3 today
            There is a draft and ready for WGLC
                Last call after this meeting
    New Work Proposals
        Naming/Addressing
        Registry of Service IDs
        Neighbor Discovery
        Routing
        DTN Key Admin

cRT:
    Naming/Addressing
        defines format
        complexity in reading between lines
        Locator/ID seperation
        How do we locate them if ID is ID?
        Global?
        Will IPN have a unique names?
        IANA, how to hand out?
        EID cna refer to multicast entity
        Lots of things
    Registry of Service IDs
        Ex. Port 22 == SSH
        Do we have common DTN services we wish to refresh?
        IPN has concept of service number, DTN does not
Marc Blanchet:
    We do this Service Registry work once BPbis is out
    Resume that work
    Very important
    Need to identity what to expect to receive where
cRT:
    Consider yourself voluntereed!
    Neighbor Discovery
        BP Discovery (ARPish)
        Service/Endpoint Discovery
        Convergence Layer Discovery
            Node IDs?
        Hybrid?
            It all just works
        This is all more than 1 draft
Scott B:
    Endorse importance of this
    ... Excellent starting point
    Core at convergence layer
Brain S:
    Did play with this in MANET
    UDPCL update would be multicast disovery to IP Layer discovery
    BP does stuff MANET does lower?
cRT:
    Looks good, nice
    Routing
        Really hard, where or when to bundles
        SABR, PROPHET
        Haven't solved, need ideas
        If starting tomorrow, beyond our current bandwidth
        Proposal: divide into folding then routing problem
        Fowarding Spec
            Routing info base and sum of clever stuff into FIB of host
            Try to define the FIB (minimum for DTN)
                This might be achievable!
            How does FIB work with other topics
Scott B:
    Think correct - still actively being researched
    Endorse idea of studying and pay attention to it but not commit to target
    Multicast for forwarding and routing and naming/addressing
Ronald Velt:
    Re routing: one important categorization is single bundle copy versus
    multiple bundle copies strategies
cRT:
    Multicasting touches a lot of things
Scott:
    Lots of things here closely intertwined
cRT:
    Interim for naming/addressing to see if any agreement
cEB:
    Routing in general for current charter and do after naming/addessing and
    mulitcast
cRT:
    Can't deal with all these without pinning down a general concept for
    filling FIB with something How its filled out of scope but if it is filled
    how do we use it - in scope and need to define it DTN Key Admin
        Scott with interesting ways to distrubute
        Run this over BP
Martin Duke:
    Clarify scope: how forwarding decision made
RT:
    If you define FIB you must define how to use otherwise just table
MD:
    Reasonable
RT:
    "How table is populated is currently out of scope"
    Information doc as reference point for other documents
MD:
    Cross area discussion in charter
Scott:
    I like idea of info doc for FIB or terms and stuff
Zahed:
    Mix of tech and charter
    Help understand??
Ronald:
    Does FIB presuppose that there is a single "next hop"?
MD:
    I hope not Ronald.
cRT:
    What does WG want to do, ordered by interest and technical need.
    Tackle naming, neighors and services, etc.
    Is there interest from others?
Ronald:
    Per EID I mean.
cEB:
    Look at BPv6 and remember that deployments and turning it into BPv7
    Work through the backlog
Zahed:
    Looked through charter
MD:
    (wearing my hat as irresponsible AD) beyond "what does the WG want to work
    on?" the other question is "what are people ready to deploy?"
cRT:
    111 propose charter?
MD:
    ^^^

.... (missed)

Stu Card:
    I keep trying to find time to work on SMTP and NORM CLs but so far no joy.
Ronny Bull:
    I am building out a robust EMANE cluster and would like to use it to help
    test and further progress the development and acceptance of BIBE.
MD:
    Link to the TAPS talk: https://youtu.be/RvFy-FFd5CU?t=627
Marc Blancet:
    DAT: DTN address translation
cEB:
    Plan intermin meeting? Any support for this?
Mike English:
    Is there use case / scoping work that could help inform future routing
    considerations?
Marc Blanchet:
    Yes to interim if one topic
cRT:
    Probably interim would be about naming/addressing
Stu Card:
    Support topic for interim
cEB:
    We must get correct
cRT:
    Flurry of email on adoption stuff

- Any other business / Open Mic, 10 mins