March 29, 15:20PM – 16:30PM, Room 253, DECADE meeting @ IETF 83, Paris
Chairs: Richard Woundy (Rich Woundy), and Haibin Song
Meeting notes taken by Richard Alimi (Rich Alimi)
Responsible area director is Martin Stiemerling
Chairs Slides (Rich Woundy and Haibin Song)
● Lessons Learned
○ Yingjie Gu: Can we use the term “DECADE-compatible”?
■Need to be clear on what it means
○ Akbar Rahman: “DECADE-compatible system would have clients and servers that talk to each other” but describe properties of the system
○ Martin Stiemerling (AD): Sounds like marketing discussion. We can just comment that it is DECADE, but it isn’t a DECADE protocol yet
○ Rich Woundy: Leave rest to offline discussion with Martin Stiemerling
■Integration examples using a prototype -- the prototype is not DECADE since we haven’t event decided the protocol
Problem Statement (Haibin Song)
● Haibin Song: Is reference to RFC3414 sufficient? Or should we change it to RFC4949?
● Rich Woundy: Do ADs care which one?
Leave it as-is
● Haibin Song: any objections to removing examples referring to PPLive or Octoshape?
● Add security risks?
● Leif: beginning had text along lines of end-to-end encryption would not work with decade. Long list of threat vectors typically addressed by encryption. Expected to see text on how to resolve conflict. Draft needs to discuss this.
Leif can help put together text
● References for P2P Caching
● Rich Alimi: Can we reference survey?
Chairs: yes that should work
● Rich Woundy: Have authors and Rich Woundy work out RFC editor notes
Send to Martin Stiemerling by April 20
● Mandatory protocol
● Haibin Song: discussed with Martin Stiemerling
○ conclusion was to not include in architecture document
● Martin Stiemerling (AD): Correct - don’t specify this in architecture document
● Akbar: Yes - that should probably be in next step of WG
● Rich Woundy: Does decade reqs say we need a mandatory one?
Yes (quotes draft-ietf-decade-reqs-06)
● Rich Woundy: Have it done by April 20?
● Akbar Rahman/Rich Alimi/Dirk Kutscher: Should be doable
Integration Examples (Ning Zong)
● Yingjie Gu: why use INS instead of DECADE?
● Ning Zong: implementation is not “DECADE” since it is still under development
● Haibin Song: No DECADE protocol exists yet
● Rich Woundy: after DECADE is defined, experiments defined in the draft may or may not be applicable to that protocol
● Yingjie Gu: Storage server seems confusing
● Ning Zong: Suggest one?
Rich Woundy: We can come up
with one offline
● Rich Woundy: Reviewers: Akbar, Rich Woundy, Rich Alimi
○ Reviews done by April 20
HTTP-based DECADE Resource Protocol (Danhua)
Rich Woundy: reminder: this is
not chartered work, but can serve as example for future work
● Remote Get message
● Rich Woundy: should explore using GET operation, and make the local DECADE server act like a proxy
● Rich Alimi: Agree - try to use HTTP GET
Haibin Song: Agree as well
● Access and Resource Control
● Haibin Song: Don’t think Kerberos is the right solution for resource control; OAuth may apply if DRP is HTTP and especially data transport
Seems like about time to just write a protocol using OAuth
and see where we run into trouble
Charter Revisions (Rich Woundy and Haibin Song)
● Rich Alimi: Be sure we’re in sync with having a mechanism other than extending data transport protocol itself
● Rich Woundy: Yes - need to see since changing charter requires more process outside of WG
● Martin Stiemerling: No comment so far - still need to digest it