Skip to Content.
Sympa Menu

md-distro - Re: [md-distro] avoiding dynamic metadata queries

Subject: Metadata Distribution Subcommittee of TAC

List archive

Re: [md-distro] avoiding dynamic metadata queries


Chronological Thread 
  • From: Ian Young <>
  • To:
  • Subject: Re: [md-distro] avoiding dynamic metadata queries
  • Date: Mon, 19 Aug 2013 15:36:39 +0100


On 16 Aug 2013, at 20:56, Tom Scavo
<>
wrote:

> Okay, so I should have said: an MDX server that supports arbitrary "ad
> hoc" queries (to use Scott's terminology) is a disaster waiting to
> happen.

I would agree with that assessment. eduGAIN isn't one of those, though.

> I think we can avoid that and still offer some significant
> capability.

I agree.

> I can't speak for anyone else but it definitely *is* what I want to do
> since it fits my world view perfectly. That said, I'm not sure it
> needs to be represented in the spec at all.

OK, that simplifies things somewhat. This would be the wrong place to
discuss extensions to the metadata query spec anyway.

> So let me summarize my main points:
>
> - An MDX server can avoid arbitrary "ad hoc" queries and still deliver
> significant functionality.

Definitely.

> - Can we agree that most everything of interest (> 80%) can be known
> in advance and can therefore be pre-computed?

For the use cases we've identified as ones we want to support so far, seems
more like 100%.

> - Let users create entity attributes and thereby pre-register their
> queries (so that the target aggregate can be pre-computed).

It could be entity attributes, or it could in principle be some other
selection mechanism. The key part is that users pre-register their queries
so that the response can be pre-computed and then served up from a can by the
back end.

It also makes charging people who use the custom aggregate service a bit
easier, no idea if that's on the cards ;-)

> - An online signing key (secured in a trusted HSM) is a separate
> issue.

Yes.

> From a security PoV, what we lose by exposing the signing key
> in an HSM might be gained by tightening the validity window and
> eliminating the human factor.

"Security" isn't scalar; you can't just trade off like that. You need to
decide which things represent actual threats you care about, and assess your
defences against them independently.

-- Ian



Attachment: smime.p7s
Description: S/MIME cryptographic signature




Archive powered by MHonArc 2.6.16.

Top of Page