ICNRG Interim Meeting 2018-09-24
09:30 Meeting starts
17:30 Meeting ends
Meeting notes: (Question and answer discussions)
1) An Analysis of secure Scuttlebutt as an ICN system - Christian Tschudin
-The Decentralised web summit (jul/aug 2018)
-Zooko's Triangle (Zcash CEO)- Decentralised web
-Replicated logs and subjective Readers : recording events on social media via log and replicate the friends log.
-SSB- Social graph-based connectivity; on-boarding problem; merits ( uses ED25519 elliptic curve crypto- one key for all ), e2e encryption - fully preserving handling of meta-data, gossip, delay-tolerant, resilient)
-RPC- secure hand shake (SHS)
-comparison with ICN
-->Conclusions: Zero need for a intermediary like FB, Twitter,Gmail etc.
secure, scalable, resilient, delay-tolerant, privacy-preserving ICN.
replicated append-only logs
data growth frontier
-->QnA: (Q) is the centralised web focused on social media /E-commerce---
(A) social media
(Q) Decentralised doesn’t mean no one is given a name: therefore Zooko' s Triangle has a logical problem and name is not a crypto-
(A): this is what the web people find it attractive about the meaning ful decentralised names
(Q)how to make new log-
(A) its totally independent to make new ids
(Q) Is access given to other users of the log?
(Q) how to provide access to different users of the log you want to share with without having different copies ?
(A) replicated append- only logs (link, replicate syn property).
(Q)there is no trust in here?
(A)there is a implicit trust in here (binding together with lot of other properties)
(Q)when public key is your identifier (a friend of friend and so on, is it visible to a friend of friend) ---
(A) its a two way process flow and follow back, being friends are two different things.
(Q) Who can you trust? just a friend or a friend of friend ? that is a one hope model
(A)there is a physical graph and the social graph that is not to be mixed up.
(Q) need for identity based encryption.
(Q) will caching work here?
(A)gossip it self is a cache
(Q) then isn’t it like the current id based system? so the channel is secure and not the data?
(A) now it ishuman centric futures works on involving technology.
(Q) what about the scaleability? to talk to the a trusted system directly without an intermediate link (range and the network is the concern)
(A) still in progress, requirement of new design.
2) Update on implementation of ICN Video distribution - Martin & Marie-Jose
8X labs (Startup)
goal: providing innovative software sol. to support video services
challenge: bring Content Distribution to the edge
tools: ICN and Smart Contracts (on the blockchain) the strength : Combine ICN + Blockchain
Montreal's interim (July 2018)
partnering with ENCQOR so that ICN is implemented in real environment
Issues: business challenge-- identify major expenses and revenue streams
ease of integration
ICN open source and community support.
(Q): reason for the smart contract
(Q): is cost saving the advantage?
(Q): parallel deployment
(A):together with the legacy system.
Challenges of a commercial implementation of an ICN for video distribution - Martin & Marie Jose
-a lot of people do not know about this and its confusing
-training under grads and grads
-more handson during tutorials.
3) Naming discussions
- ICN - definition- location independence
-internet conflates location and identity.
-Host Centric :Hip, Information Centric : TRIAD
-different approaches to Location Independence
- location and mobility (multi homing) (important one but its not the only one)
- mobile IP, Indirection routing ,name to address resolution, name based routing (best).
-if the home node is down then the start of communication is difficult
mobility first:mobility-centric+trustworthy-- GUID pre looked, connect-time, individual, simultaneous mobility
scaleable global names services
DNS limitation: Passive Caching, static placement, hierarchy .
GNS: Decoupling certification and resolution.
Context based communication.
scalability: impossibility of linear scaling.
-(Q)node (anything) security
-(Q)so it is not name specific? the name is location independence and replicated
-(A) its just the name it may be a content , node or just anything. (just using the name space), its only mobility and not multihoming
-(Q) looks like there is no third part but in real there is one
-(A) trust is always a part
(Q) centralised trust and distributed trust
(Q) TLS is stronger in browser?
(A) lack of deployment of GNS - DNS is not secure already
(Q) name collisions?
(A) CNN -china news network, CNN news channel?? so there is a confusion. (conflict) and this requires a trust entity (at least one)
(Q) do u see ICN and mobility first combine
(A) mobility first is just another ICN.
We break for Lunch
4) Update on ICN-Lowpan work
-ICN- Lowpan Architecture
-current implementation state CCN lite.
- Q)what is the relationship between hop id and face id
-(A)face is the logical interface, flow id on a per hop bases.
5) NDN measurement framework and NDN Traceroute tool
(Q) how many are using the collected measurement data?
(A) might need the data or might not need the data., ("when "is the parameter used to extract when you want it to be done) ; or send the data to the repository and the client retrieves the data from there
(Q) the idea about how to design the name space.
6)NDN IoT Packaging - plan to release in October 2018.
Learning: Documentation, Developing community,
IoT Device Software Framework, Bootstrapping, Secert public key, lightweight access control, Adaptation layer,
(Q) what n/w interface is this implemented on?
7) Deep dive into ICN & QoS
Thoughts on Quality of Service for NDN/CCN
QoS:- managed unfairness, doesn’t include billing, QoS in TCP/IP-link capacity, Router Forwarding capacity.
why is ICN better: Hierarchical Names are a much richer basis for specifying equivalence classes than IP 5-Tuples; Intserv requires flow signaling with state O(# flows), three forwarding semantics;
need to do something clever with ICN related to QoS. IP treats all endpoint as open-loop packet sources.
Ip has no Caching;
A Strawman set of Principles.
(Q) in valid traffic at times QoS is really important to retrieve data.
(Q) what if the data is of variable size
(A) the counts are by packets and not bytes
(Q) say larger packets as small and small as large?
(A) depends on the design of the algorithm.
(Q) what is the data size is specified?
(A) can’t specify.
(Q) who is blame if there is no QoS?
(Q) In ICN we have no bits defined whereas in IP we have 6.
you cannot take a network that doesn’t work and add QoS and make it work.
Conclusions: planning for next ICNRG meetings at IETF 103 in Bangkok
" its all about the travel in this experimental path"