DINRG Meeting @ IETF-104, Prague

Note taker: Brook Schofield

Logistics

09:00 Welcome, Agenda Bashing - Chairs
09:10 Transport Issues for End-System Multicast and Message Propagation in Distributed Ledger Technologies - David Mazières
09:40 GNU Name System - Christian Grothoff
10:10 Byzantine Agreement Protocols for Large-Scale Decentralized Identity Management - Nathan Aw (remote)
10:40 NGI Trust Open Calls - Brook Schofield
10:50 Wrap-up -Chairs
11:00 End of Meeting

provided a summary of how the bitcoin network works
Melanie: How are you going to take this work forward?
David: Looking for a set of requirements that can be taken forward for research.
Maria/MIT: Based on a peer-to-peer video distribution system there would be some security from some FEC headers. It was a coded version of DHT.
David: The type of "security" you are talking about is confidentiality
Maria:
Tobias Cuc: Have you considered key distribution techniques adding confidentialitiy
David: Again confidentialitity isn't need a requirement but availability is important. It would be important to get use cases or examples where this was needed.
Tobias: Message Layer Security (MLS)
Kurt: Have you looked at password sync and data reconciliation and signing of messages?

David: Can you add about adding randomness for eclipse attacks against Kamadelia.
Christian: This doesn't solve all civil attacks.
Shabira: What is the information is in the key when the active node in the cluster fails?
CHristian: The information is published into the DHT so it is cached for some time.
Giovani: Excited about new name systems but the real thing about DNS is high availability and performance.
Christian: DHT can do caching and also at the client side to improve performance. Caches can prepopulate because you know the time that a label will expire in the future.
Giovani: How could GNS work for more than 100 million domain names?
Christian: Most of the churn of DNS records comes with DNSSEC and that isn't needed within GNS. Rapid IP address cycling does take a performance hit.
Ulrich: How do I first get the ietf.org GNS key to do my query?
Christian: You get the key from .org and use that to lookup ietf.org so that would be prepopulated.
Wes: Really cool math behind this and the TLSA screenshot from your presentation. You're collecting keys and you don't have a global name so how do you know whether the key is served via global concensus or local configuration?
Christian: It is developed via global concensus.
Ace: What happens if .rms (in your example) becomes a global name via DNS you'll have conflicts.
Christian: We recommend that ICANN doesn't issue any more global names.

David: Regarding the IPR policy please note that alogrant?(spelling) is patented.