I've made an attempt at rewriting the Cache section.  I've put a table of operations at the start, clarified that there are potentially many cache services (one per set of interfaces), expanded the description of the create operation, added a list of properties, and added sub-headings to improve the structure.  I've borrowed a lot from the replication section. 
 
Mario, if you let me know when you have a suitable version, I'll copy these changes into it.  I can also update some of the glossary entries.
 
At some point I need to work on the Catalog section.  I'll try to do this soon.
 
Other thoughts:
 
In my view, section 2 should give an overview of the components of our toolkit, the capabilities they provide, some high-level patterns for their use, and some cross-cutting concerns.  For example, it should state that we describe storage management, explain the interaction of access and transfer, state that replication can be used to control these, and that composite services can be created using federation and caching.  It should explain how these services may be hidden behind client libraries (e.g. for legacy support).  Finally, it should describe how policy issues are handled across the various components.
 
Section 3.7 should be discussing the support for policies provided by OGSA, e.g. WS-Policy, WS-Agreement, etc.  It's not our job to define general mechanisms such as these.  Currently this section does talk about data-specific issues as well.
 
I wonder whether we should say something in section 2 about service creation.  Throughout our document we have tended to assume the existence of factory services, each with an appropriate createXXX operation.  I think this is a useful way to structure our document; the createXXX operations let us describe the issues that need to be taken into account when creating a particular service (such as policies, location, etc.).   But in the larger OGSA context, might services be created as part of the provisioning capabilities of the architecture rather than a specific factory service?
 
Section 10.3 still discusses a DestroyReplica operation.  I think we agreed to remove this (it's not in the table at the start of section 10).
 
 
Best wishes,
 
Dave.
-----Original Message-----
From: ogsa-d-wg-bounces@ogf.org [mailto:ogsa-d-wg-bounces@ogf.org] On Behalf Of Allen Luniewski
Sent: 08 June 2007 16:31
To: ogsa-d-wg@ggf.org
Subject: [OGSA-D-WG] Teleconference Minutes, 8 June 2007


The minutes of today's teleconference have been stored on GridForge.

Please note that Mario has the write token on the Data Architecture document.

The date of the next teleconference is TBD but most probably the week of 25 June.

Allen Luniewski
IBM Cross Brand Services
IBM Silicon Valley Laboratory
555 Bailey Ave.
San Jose, CA 95141

408-463-2255
408-930-1844 (mobile)