Skip to main content

Minutes IETF114: cdni

Meeting Minutes Content Delivery Networks Interconnection (cdni) WG
Date and time 2022-07-29 16:30
Title Minutes IETF114: cdni
State Active
Other versions plain text
Last updated 2022-08-01

CDNI WG Minutes
IETF-114 Philadelphia
Chairs: Kevin Ma and Sanjay Mishra
AD: Francesca Palombini

Agenda and Slides:

Chair Slides (Chairs)
RFC9246 published!
SubCerts draft adopted!

CDNI Footprints (Sanjay Mishra)
- During WGLC 2 issues were raised:
1. Should we use earth based coordinate areas?
  Sanjay: subdivision code solves a specific known use case requirement; the
  proposal: if a requirement arrises write a separate draft Kevin: Agree with
  the proposal Chris Lemmons: Agree with the approach.  The implementation
  could be gnarly.  Wait until it is really important. Benson Muite (from
  jabber): That is ok
2. Should we register the ALTO Entity Domain Types in this draft or a separate
  Sanjay: Will review with Nir to pick the best option
  Kevin: Happy either way; would like to get the draft finished up
  Emile Stephan: Just add it to the IANA section of this draft.  Creating a
  separate draft for the ALTO is more work.  It is more efficient to just add
  it to this draft Kevin: It shouldn't be a lot of text

CDNI Triggers (Sanjay Mishra)
- The trigger extensions draft added new features; 8007bis changed encoding for
extensibility - trigger spec: generic object - trigger type -> generic action
Sanjay: Authors will do editorial updates and would like to get to WGLC for
IETF 115 Kevin: Everyone should review the changes

HTTPS Delegation (Frederic Fieau)
Frederic: Draft is updated.  Are we ready for WGLC?
Kevin: I sent more comments to the list this morning: not sure if we need the
FCI object or metadata explanation and may need an update to the security
considerations Sanjay: Thomas Fossati is reviewing the draft as well Chris:
(channeling Rajeev RK in jabber) Slide footer should not be "Orange Restricted"
Frederic: Will update Kevin: I would like to get to WGLC before IETF 115 if we
can address the comments on the list

HTTPS Delegation using subcerts (Christoph Neumann)
- Updated the name to be a working group draft.
- Removed the new interface and added an FCI object
Kevin: I emailed list this morning: can we get away with not using FCI and just
using metadata?  This is an important topic that we need to close on.  Everyone
should provide input on the list.

Sanjay: I also sent comments to the list this morning
Christoph: The main point is having the FCI object discussion.  Will follow up
on the list. Kevin: Yes.  I will hold off on nits until we close on FCI vs
Metadata changes.

CDNI Metadata (Glenn Goldstein)
- SVA has a push-based metadata API: need to distinguish update and deploy
Kevin: Is there a reason the CDNI triggered pull approach was not sufficient?
Glenn: Content Provider configuring CDNs generally push
Kevin: Right, and CP -> uCDN is out-of-scope of CDNI, so we didn't do CP ->
uCDN push - Proposal: 6 smaller drafts, with new content Kevin: I like the
smaller, scoped drafts.  For the new content, encourage having the discussion
on the CDNI list. Chris: We should verify charter scope for each draft Chris:
Significant support and excitement for metadata expression language Tom Hill:
Why not use YANG for metadata expression language?  Is there room for YANG? 
For automation, many are familiar with YANG. Glenn: I can't speak to YANG.  The
metadata expression language is based on internal work done at Lumen, but we
will follow up. Chris: There is less YANG experience in the WG Emile: There are
many languages <technical_difficulty> Alfonso Silóniz: I'm not familiar with
YANG, but it is a good thing to discuss Glenn: YANG is a data modeling language
and we need an  expression language which is different Alfonso: Agreed Ben
Rosenblum: We use YANG for configuring telecom hardware Rajeev: There is also
an expression language syntax that might be usable without using the modeling
language </technical_difficulty> Glenn: We will address YANG in the list thread
- For new FCI objects, should they live with the metadata definition? Kevin:
Keep them together Glenn: Agreed - Name footprints RajeevRK: Footprints are
defined by dCDNs but there is no way to otherwise reference them Kevin: Could
we just add a label to footprintunion in the footprint draft RajeevRK: Will
read the draft RajeevRK: Is it advisable to have overlapping footprints? Kevin:
CDNI does not have a stance on overlapping footprints; it's up to the CDNs to
determine how to use them Matt Stock: Composite footprints can become hairy.  I
like having an optional name.  Or a way to ask the dCDN if it can support a
footprint X and have it return a label. Alfonso: If a dCDN advertises
capabilities with different footprints, it is hard to manage metadata
configuration because it is not possible to define metadata by footprint Kevin:
We have not looked at this level of complexity before.  It is a big topic that
we should take to the list. Glenn: Related: we have also looked at named

Consumer Technology Association (CTA) Web Application Video Ecosystem (WAVE)
(Chris Lemmons)
- more concrete list of claims
RajeevRK: Network claim includes ASN (not just IPs)
Richard Patterson: What happens rotating between IPv4 and IPv6 or when privacy
extensions are used? Chris: They don't work and the doc says don't use this for
identity binding.  There are some legitimate uses (e.g., allow only 10 net)
that demand support. RajeevRK: IPs more granular than /24, /56 must be
encrypted - encrypted composition claims Kevin: What is the impact to CDNI?
Chris: Not a direct competitor to URI Signing.  May need a draft to support
delegating CATs; metadata for CATs.

Capacity Advertisement (Chairs)
Kevin: Does anyone think it is a bad idea to adopt the draft?
Chris: :thumbs_up:
Ben: Andrew has a new revision with cosmetic changes if we want to wait to adopt
Kevin: We can do the call for adoption.  I encourage everyone to read the draft
and provide input on the thread on the list.

session closed