Interfederation

[Tom Scavo <>]

On Thu, May 2, 2013 at 9:50 AM, Jim Basney 
<>
 wrote:
> I added a new section at
> https://spaces.internet2.edu/display/incinterfed/Interfederation+Lessons+Learned
> about opt-out / opt-in to try to capture the group discussions. Feel
> free to edit. We'll pick up the discussion on our next call, but in the
> mean time please don't hesitate to continue discussing via the mailing list.

I understand Scott's viewpoint. He's essentially said the same thing
many times in the past. I don't disagree.

(As an aside, about half of the R&S SPs expose all InCommon IdPs on
their discovery interface. One in particular, GENI, is driving IdPs to
support R&S. Every time Tom Mitchell sees a federated transaction
fail, both of us gang up on the IdP in question. Together we've
transitioned many IdPs to R&S. I wish all R&S SPs did the same thing.)

Applying Scott's general strategy to the LIGO use case, why not do the
following:

- The LIGO SP directly consumes a UKF metadata aggregate (like it does now)
- The Cardiff IdP directly consumes the InCommon metadata aggregate
(no special "export aggregate" needed)

And we're done. I can imagine some minor optimizations to the above
model but that's basically it.

I'm not suggesting this lightly. I'm asking a real question that has
bothered me for some time. What value can InC Ops add to the UKF
aggregate? Likewise, if we adopt Scott's all-inclusive point of view,
why would InC Ops want to create a special "export aggregate" for
interfederation purposes?

Tom