[GSM-WG] SRM (GSM) and OGF

[Jensen, J (Jens)]

Folks,

Since another OGF is coming up, we should think about how to make 
progress with the standardisation.

Since we now have a good five or six interoperating SRMs, it is perhaps 
time to revisit the SRM 2.2 specification.  The SRM 2.2 doc really dates 
back to Aug '06, and people are likely to have deviated from it in the 
implementations.  This is quite natural: only when implementing a 
protocol do you really grok it.

I would propose we start an activity to review the spec and document at 
least the most obvious discrepancies.

The aim is partly to ensure that someone new who picks up the standard 
can start implementing an interoperable SRM.  Partly that we need to be 
responsible standards maintainers, now that SRM is an OGF standard.  Of 
course if there are no changes then so much the better, but it would be 
good to be able to say so.

Secondly, I suggest it would be worth while to start collecting open 
questions.  If we have enough issues that need addressing, perhaps we 
can have a revision 2.3 of the SRM protocol.  For example, how subspaces 
behave - can they be reserved?  Supporting non-SOAP APIs?  Which 
function calls are not implemented, such as request suspension? 
Querying spaces that you own?  Use of space tokens while reading?

In particular, several people have requested changing the delegation 
mechanism to not use GSI delegation.  gLite has a delegation API which 
we could use instead.  We could carry SOAP messages over plain SSL/TLS 
which would simplify many things.

So there you have it.  Suggestions and comments welcome.

This mail was sent to gsm-wg and cc'ed to srm-devel - in the interest of 
openness, we should continue the discussion on gsm-wg.

Thanks,
			--jens