Comments on architecture draft

From: Andrew Lawrence <al-at-roe.ac.uk>
Date: Thu, 29 Apr 2004 17:18:46 +0100


Dear all

I have been reading through Roy's architecture document. Overall I thought it was really good, and very useful. But its not really doing the job I was looking for.

This job was really my fault. Before the last IVOA meeting, AstroGrid folk proposed a "Technical Architecture Group" and new working groups in Virtual Storage, and in auth/auth. This was controversial, and I proposed a temporary "Architecture Tiger Team", with the idea that this review would survey the landscape, identify the blank areas on the map, and send out explorer teams. In other words, the Task Force would decide whether we needed any new Working Groups. A gap analysis, if you will. Roy's document is an excellent statement of the current state of play, but does not address what else we need. (Although MySpace, auth/auth, and Workflow are all mentioned in the introduction, and present in the diagram.) On the other, it lists two new working groups, one of which isn't a working group (Applications : should be an "interest group") and one of which has come out of the blue ("Data Engineering").

My suggestions :

(i) Explain the purpose at the top. Make it clear its not a Technical
Architecture (IVOA is not a system) but is a "Conceptual Architecture" or "Landscape" or suchlike, in the context of which we can develop standards and best practice.

(ii) Purpose number one is a survey of the landscape at this point in
time.

(iii) Purpose number two is to do a gap analysis and make
recommendations on any new needed Working Groups.

(iv) The section labelled "Architecture Components" is really "Current
Activities centred on the needs of Architecture Components". The components have been pretty
much listed in "Overview". This section should remove the "Data Engineering" and "Applications" section. Then there could be a section labelled "Possible new working groups" with short decsriptions of Data Engineering, workflow, auth/auth, virtual storage, and "semantics" or similar. Finally then a section called "Interest Groups" with short descriptions of Applications Interest Group and Theory Interest Group.

(v) Then we need a "Reccommendations" section ....

Other points :

End of section 2 : "these services will be implemented in terms of Grid middleware functionality.." As long as we interpret "Grid" loosely I guess this is ok... For example we are developing auth/auth using a simple system that is in the spirit of GSI, but absolutely does NOT require users to go to a proper Grid certificate authority. (The AstroGrid system is based around IVORNs)

"The IVO is collaborating with a number of IT groups that are developing workflow software". Errmm well actually we are writing workflow software. Correction, we have written workflow software. It works. And we had better make sure that we can exchange workflow files ...

"A vital part of the IVO architecture is the concept of MySpace". If its vital, sounds like we need a working group....

Data Access Layer : "We must also include some client side software...ultimately most software will come from .." This is a bit out of place. Who is "we" ? IVOA doesn't write any software. It sets standards and debates best practice.

UCDs. I haven't kept properly up to date with this WG, and don't understand "UCD3" but is it not controversial whether proper semantics or ontology or whatever is really UCD++, or is superseded by Data Model, or is really something new ? What I am getting at is maybe we need to envisage the semantics IG being promoted to a WG ? Or maybe its still too early ?

       andy



==
Andy Lawrence                                 e-mail      al-at-roe.ac.uk
Head of School of Physics                     phone (UK)-(0)131-650-5247
University of Edinburgh                       http://www.ph.ed.ac.uk/
PA : Doreen Hamilton                          email  D.Hamilton-at-ed.ac.uk
                                              phone (UK)-(0)131-650-5249
                                              fax   (UK)-(0)131-650-5902
----------------------------------------------------------------------
Regius Professor of Astronomy                 email       al-at-roe.ac.uk
Institute for Astronomy                       phone (UK)-(0)131-668-8346
Royal Observatory Edinburgh                   http://www.roe.ac.uk
IfA Administrator : Liz Gibson                email eg-at-roe.ac.uk
                                              phone (UK)-(0)131-668-8356
                                              fax   (UK)-(0)131-668-8416
========================================================================
== Received on 2004-04-29Z16:19:19