IETF 113 DTN working group meeting

Tuesday 2022-03-22 13:30 - 15:30 (UTC)

Co-chairs: Rick Taylor, Ed Birrane
Secretary: Adam Wiethuechter

Abbreviations:
AD Zaheduzzaman Sarker - adZS
Chair Rick Taylor - cRT
Chair Ed Birrane - cEB
Adam Wiethuechter - AW
Vinton Cerf - VC
Scott Burleigh - SB
Brian Sipos - BS
Emery Annis - EA
Sarah Helble - SH
Rick Taylor - RT
Ed Birrane - EB
Jorge Amodio - JA
Joshua Deaton - JD
Ronald in 't Velt - RV

Admin, Chairs, 5 mins.

The State of DTN (25 mins): Vinton Cerf

• Thanks for inviting!
• Congrats on your progress!
• Many folks around the world about the work you are doing and opportunity to demo the standards
• IPNSIG changing to Interplanetary Networking Chapter
• About 800 members scattered around the world
• Working groups
  • Strategy: released stuff 2021 avaible on website; 100 years ahead
  • Library: documents
  • Business: trying to understand the commercialization of space.
• Strategy Report of what could happen next 100 years

Intentions

• Test DTN in terrestiral and LEO configurations
  • Prepare for deep space
• Network management
  • Not the same as normal networks
• Test applications
  • Messaging, AI/ML, video, etc.
  • Encourage bringing forward other apps
• Multi-network construct functional
• Collaborating with many groups to achieve testing
  • Space Agencies; preparing for return to the Moon
• Governance of operations in space
  • Mining, space labs, etc.
  • Artemis Accords
    • What does it mean to own private property on Moon?
• Public attention
  • Signifcant agenda been adopted by IPNSIG
• Questions or corrections?

Q and A

EB: lots of people present; as mind behind tech give a few sentences why?
VC: remember in '97 path finder was landed on Mars after many failures. Following spring met team for this path finder, what is needed 25 years from now? Need internet for solar system. Why can't use TCP/IP? Found flow control breaks due to distance/time. And thus bundle protocol began development. Need resilience due to physics limitations.
adZS: Thanks! Look really cool. a) for IETF WG and rechartering, have any expectations?
VC: Continue to refine protocols and take into account we have to deal with delay but work in real time. With low latency do streaming video. Apps function in both present day internet and on DTN. More energy into developing DTN applications. Naming and addressing strucutres to regularize and how to manage. IANA and SANA. Commercialization; what does it mean to use in a mixed envirorment (government, private)?
adZS: need broader ietf interaction, thank you!
VC: thanks and wish you well!

cRT: I do have slides on a couple of topics which he pointed out.
cEB: we are sharing slides for our presenters

JA: As you already know DTN/BP has been on the drawing board for a while, it is about time (not talking about Bitnet) to finalize testing, get reliable running code, standarize the protocol stack, and start bulding applications for both space and terrestrial use cases. On the other hand, my actual Testing Lab project Vint mentioned in his presentation is about educating a new generation of young students inspired by STEM programs, they will be in years to come the end users of this technology.

BIBE and Quality of Service (25 mins): Scott Burleigh

BIBE

• Very little change in Spring 2021. More refresher
• Starts in 2009 from MITRE
• Support for Content Networking in bundles and defense against traffic analysis
• Missing a lot of things until 2013
• Not as application on top but as a CL underneath
• Discussion arount custody transfer
  • Some problems: retransmission was around RTT. Fragmentation would break the transfer
• Concept: BP transmission reliablity needs to be between neighboring nodes
  • Just add custody transfer into BIBE as fits concept
• Aggregate Custody Signalling
  • Optionally reliable CL under BP
• Slide 10 - single slide overview
• What if accurate clocks missing?
  • Either side of the link
• BIBE can be helpful for some weird scenarios
  • Cross-domain security
  • traffic analysis
  • transient quality of service
  • critical forwarding
  • multicast

RT: slide 12; parties not having clocks? Both don't have synced wall clock or cant count seconds
BS: Its a broken clock
RT: can count time
BS: can only do in interval
JD: BIBE considering bring to CT together? or is it an extension?
cRT: not tied together in charter but understood as related. lets do that at the end as in depth.

Quality of Service

• Concepts
  • BP job is to foward bundles over topology to adjacent nodes
  • when queue of bundles; which forward next?
  • QoS is alternative algo for this question on queue
• Survey
  • Overprovisioning
  • QoS in Internet
    • Intserve
      • Reservations of sender/rcvr pairs
      • Doesn't scale
      • Not delay tolerant
    • Diffserve
      • Closer fit
      • DSCP in header for per-hop behavior
  • BPv6
    • Class of service
      • Like mail, standard and express
      • Not defined
      • Does not cost anything to mark as express, its an honor system
    • Extended class of service
      • Prototyed in ION implementation
      • ordinal tag
      • select flags
• Granularity
  • No head of line blocking
  • Yes for IP
  • BP no as bundles can be big
• Suggestions
  • Add flag to request for qos handling
  • QoS Block?
  • IANA BP labels for per-hop behavior

Q and A

RT: going in the right direction!
cRT: can we do CT without BIBE?
SB: only makes sense to do it as a reliable CL; hence why removed from BPv6. Why have two when can do one for both?
RT: more effiency on wire due to bundling extra things
SB: every CL wraps
adZS: No guaratees like diffserv. Need to think more before mimicing diffserv as it wasn't successful
SB: diffserv is reasonable idea but not the model for BP

cRT: need some interims on these topics?

BPv7 Administrative Records, COSE Context, UDPCL (20 mins): Brian Sipos

Admin Records

• RFC9171 there was record types
• Add BPv7 to IANA registry
• Request adoption -- place holder for Admin Record allocations

cRT: noted, will do on list

COSE Context

• BPSec has default is intentionally limited
• Need PKI for interet facing nodes
  • PKI can be off the shelf using COSE
• COSE fits very well within BPSec
  • Adds more extensability
• One codepoint for BIB and BCB
• COSE did not define usage profile
  • More of a container for data
  • Interop profile
• Allows BPSec in PKIX environments
  • Just gets it on the wire

EB: Speaking of BPSec perspective. Fan of this work. Lots of different deployment ways. Would love to see it adopted.
BS: NASA and CCSDS looking at shared PKI environment?

Bundle Version ID

• Not well defined way to determine if BPv6 or BPv7 explicitly
• There are tools that assume BPv7 is encoded like examples
• Good news that no collision between encodings
• CBOR stream decoding can work
• Might be worth documenting some optimized ways
• Not draft yet, but is there interest?

RT: I think yes short Info for interop is good.
JD: interop with v6 v7 for use with ISS followed same idea. Started with v6, verify for v6 then v7. There is value in this.

UDPCL

SKIPPED DUE TO TIME

DTNMA/ARI (15 mins): Emery Annis

• Firmly falls under new charter
• Renamed for alignment
• Focus on challenged networks
• model driven arch
• clarity added to roles
• Questions
  • What is missing, what is wrong

EB: observation; renamed doc. Saying "asynchronous" caused confusion - we mean "transport-layer asynchronous requiring autonomy" and others see asynchronous and think "REST-ful interfaces over TCP"
JD: maybe missed; does std cover for handling reports that have not been sent for extended periods of time?
EA: no answer in std, could be accomplished separately. apply policy to them.
cRT: clock. Arch document can proceed with accompany documents. Please reread and review.

• New document (personal) to name resources in app models
• Interested?

cRT: within scope, push and draw interest. 2 sessions next ietf
cEB: agreed.

ANMS Demonstration (10 min): Sarah Helble

• First spiral complete, going through review

JA (chat): Is there any ANMS code available we can play with at IPNSIG ?
EB (chat): The ANMS team is working to open-source an alpha release shortly. Perhaps April. The idea is to gather early users who can both test the system, propose features, and be part of the support community for this. This would be release of spiral 1 of 5 planned development spirals over the next 2 years, so we would have time to gather experiences and react to them.

On Endpoint IDs (10 mins): Rick Taylor

• Tackling naming first
• BPv7 is a pair, schema and content
• IPN and DTN schema are defined
• Multiplicty of an endpoint

SB: ipn is always singleton

dtn for universal use, how to write protocol handler for a tool?

• part of IANA URI
• suggest: when used extenally (like curl) than URI is passed as 9171, th name component using DNS into IP address and TCP-CLv4 session to that address
• This is enough to use BP on public internet

cEB: posts on mail list and interim?
JD (chat): Question was going to be for verification the name being proposed to resolve to DNS is dtn://address correct?
EA (chat): Do you think the dtn:/ schema, could be extended to support things like named data networking? or would that need a new schema
SC (chat): I have some thoughts involving HF-ALE ids and DRIP DETs / HIP HHITs (which are IPv6 address compatible IDs, not locators, based on public keys).

ipn is restricted in capability, WG recommendation for usage?

• proposed for internal use as lan scope

if SDO wants its own schema how should it be done?

• who owns global registry? can we set one up?
• Go to IANA

RFC4838 Update? (5 mins): Ed Birrane

• time to refresh?

RV (chat): Previous charter said not to touch 4838

Open Mic (5 mins)