Interfederation

["Basney, Jim" <>]

We plan to meet twice in May. No calls on May 7 or May 21.

Proposed agenda for May 14 call:
* Completing review of
  
https://spaces.internet2.edu/display/incinterfed/Interfederation+Lessons+Le
arned

Proposed agenda for May 28 call:
* Discussion of new eduGAIN policy package: http://goo.gl/17Hf4

Minutes from Apr 30 call:

attending: JimB, IanY, MarkS, TomS, IJK, ScottC, ScottK, SteveC

Update on LIGO SP + UK IdP pilot:
Success with Cardiff IdP and LIGO SP!
Small change required to make artifact resolution work.
LIGO SP doesn't require artifact, but will use it if IdP metadata
advertises artifact.
Some LIGO users at Cardiff were able to authenticate to LIGO wiki!
Issue: ePPN from Cardiff is opaque. SP wiki software doesn't handle
this well. No registration process to choose good wiki name.
Currently creates wiki name automatically from ePPN.
ScottK very interested in seeing production interfed metadata aggregate.
Try Glasgow IdP next? IanY and ScottK will work on this.
Interfederation Lessons Learned discussion continued:
Process for this group is to make recommendations to TAC which
discusses with Ops & Steering.
Ops plans to add <mdrpi:PublicationInfo> top-level element.
Then adding <mdrpi:RegistrationInfo> elements per entity would be next.
REFEDS registration practice template might not arrive in a while.
Better to proceed with InCommon registration practice documentation
then fit to REFEDS template later.
eduGAIN requires only a lightweight operational statement.
The RPS statements are linked on the eduGAIN status page at
http://www.edugain.org/technical/status.php.
Quote from KALMAR participant: if we have a question, we call on phone.
Good to document registration practices, but not a blocker.
MarkS: Working through UTSystem metadata for hierarchical federation
support is a good idea. Regionals not so far along yet. UNC System
federation with about 4 of 16 members in InCommon.
Is policy involved in hierarchical federation?
UT System members have already joined InCommon.
UNC is a different case. NCTrust regional federation.
Moving forward with UT System would test technical aspects.
Could also look at consuming REEP metadata.
Policy issues will come up with regionals.
Prioritize eduGAIN or bilateral agreements?
Difficult to predict speed of eduGAIN work moving ahead.
Maybe pursue bilateral with Canadian Access Federation if eduGAIN
moves slowly?
Could InCommon join eduGAIN in next 6-7 months?
New eduGAIN policy package at http://goo.gl/17Hf4.
Definitely were concerns with old eduGAIN documents.
eduGAIN participants have now voted to accept new policy package.
Next step is for eduGAIN policy to be ratified by GEANT committee on
June 26. Then 3 month change period (though that might be waived).
UK and Austria are candidate eduGAIN members.
UK plans to sign eduGAIN declaration very soon.
AI: Jim to ask Brook if he can circulate new eduGAIN policy package.
DONE. Brook says is OK.
Note IanY can't join us May 14.
Proposed May 28 agenda: review eduGAIN policy package.
InCommon Steering moving toward more "board" function.
Good to get recommendation for eduGAIN to Steering.
Goal is for this group to recommend that InCommon join eduGAIN.
This group should review eduGAIN documents and give our
recommendation.
If a bilateral relationship comes up that will deliver significant
value to InCommon members, no reason not to pursue the bilateral
relationship (in parallel with eduGAIN - multilateral case).
InCommon and UK Interfederation roadmap discussion continued:
ScottC: Long term should be opt-out, not opt-in.
TomS: 1400 entities in InCommon. Large % not interested in interfed.
How do we distinguish between interfed candidates and others? R&S?
Export aggregate initially containing R&S SPs/IdPs?
Opt-out cases: entities like CMU local SPs, contract-based vendors.
Opt-in would limit adoption.
Rather than opt-in for IdPs, support IdPs being marked discoverable.
Need interfed agreement on discoverable tag.
eduGAIN previously mandated opt-in.
UK will do opt-in in short-term and opt-out in long-term.
Need to separately discuss SP and IdP.
Could use R&S category on IdP side in short term.
Tackle IdP discoverability in long-term.