[dais-wg] Fw: WS-DAI progress

Fri Jun 10 05:50:44 CDT 2005

Paul 

Thanks very much for the comments. Haven't gone through them in detail yet 
but will do. There is debate going on about what the conduit is all about 
so we will try and get some clearer descriptions out shortly. 

Regards

Simon

Simon Laws
IBM Hursley - Emerging Technology Services
Mail Point 137, Hursley Park, Winchester, Hampshire, SO21 2JN
Tel: Internal 248708  External +44 (0)1962 818708
Email: simon_laws at uk.ibm.com

"Paul Watson" <Paul.Watson at newcastle.ac.uk> 
09/06/2005 14:17

To
Simon Laws/UK/IBM at IBMGB, <dais-wg at ggf.org>
cc
"Susan Malaika" <malaika at us.ibm.com>, "Mario Antonioletti" 
<mario at epcc.ed.ac.uk>, <amrey at epcc.ed.ac.uk>, "Norman Paton" 
<norm at cs.man.ac.uk>, "Savas Parastatidis" 
<Savas.Parastatidis at newcastle.ac.uk>
Subject
RE: [dais-wg] Fw: WS-DAI progress

Simon,

I?ve worked through the specification?very interesting? here are some 
comments?

p5 Section 3.1. ?There will be situations where data is created in the 
context of a DAIS data service alone??. I wasn?t sure why the distinction 
was being made between this and the case described in the previous 
paragraph. In both cases a data resource is managing the data, and in both 
cases it is the data resource that is responsible for managing the 
lifetime of the data (see comments on p9).

p6. Section 3.2. I didn?t understand what a data conduit is, or why it 
needs to be introduced. From the description it sounds the same as a data 
service (e.g. to give a context within which a data resource is 
identified). Section 4.1 says that a conduit is a data resource, but that 
isn?t stated in Section 3.2.

p6. Section 3.6. To avoid confusion, I wonder if it may be better to call 
this a ?Data Resource Abstract Name?, i.e. to distinguish it from a 
service abstract name.

p6 Section 3.7. Similarly, I wonder if it may be better to call this a 
?Data Resource Address?.

p6 Section 4.1. Type 2 resources. Does it have to be a DBMS? Or could it 
be just any data management system?

p7 Section 4.1. Type 3 resources. I wasn?t sure why it was necessary to 
make a distinction between Types 2 & 3. Could we think of Type 3 data as 
being managed by some form of data management system within the DAIS 
service? If so then Type 3 is the same as Type 2.

p8. Section 4.7. ?The type and behaviour of the new database service and 
data resources are determined by the configuration parameters passed in 
with the original request.? This makes it sound as if it is 
possible/normal to create a new service with arbitrary type and behaviour. 
Is it instead more likely that the result will be made available through 
an existing, already deployed service? This does not preclude the choice 
of the service/resource combination being made by the service to which the 
request was sent, possibly taking into account information sent by the 
consumer in the request.

p9. Section 4.9. Lifetime. As with the comment on p7, Section 4.1. I 
couldn?t see why there was the need to distinguish between Type 2 and Type 
3 resources. It seems that the distinction is only in the policy of the 
data resource regarding lifetime of data. For example, it may be that for 
Type 2 data, the policy is that when the resource is removed then the data 
is also removed (because the data resource knows that the data it holds is 
only accessible through that one service+resource combination)? in which 
case the distinction between Types 2 and 3 disappears.

p14. Section 5.2.1. The message patterns look good.

p16. Section 5.3. Earlier, in the definitions part of the spec, this 
approach was (I think) referred to as indirect data access. Now it is 
described in a ?Data Factory? section. As it is not necessary for a new 
data resource to be created by indirect data access operations, and all 
operations on the service (direct or indirect) can create new data, I 
think that it would be clearer if the term "indirect data access" was 
used, rather than ?factory?.

 p15. Section 5.3. It wasn?t clear to me what was meant by ?such messages 
create a new relationship between a data resource and a data service. Also 
?In this way a data service may be used to represent the results of a 
query or act as a place holder for data to be inserted into a data 
resource.?? I was expecting this to refer to a data resource rather than a 
data service. Similarly, I therefore wasn?t sure what it meant for ?a data 
service to act as a place holder for data to be inserted into a data 
resource?.

p16. Section 5.3. ?The RowSet could be stored as a table in a relational 
database or decoupled from the database, but the important distinction 
here is that the data is represented as a collection of rows via a data 
service that does not implement the SQLAccess portType.? I wasn?t sure why 
this distinction was important.

p17 Section 5.3.1. wsdai:Configuration. See comment on p8. Section 4.7.

p17 Section 5.3.1. I was wondering if it would be useful to include some 
metatadata in the response messages from all indirect data access requests 
(e.g. data size, number of rows). This would help the consumer in making 
decisions regarding what to do with the data (e.g. how much space would be 
needed to store the data if it were fetched)

p18 Section 7.1. Will there be a ?Mapping to WS-I? section?

So, overall it looks good though I didn?t understand the need for:
a)       conduits
b)       a distinction between Type 2 & 3 services

Regards
Paul

From: owner-dais-wg at ggf.org [mailto:owner-dais-wg at ggf.org] On Behalf Of 
Simon Laws
Sent: 08 June 2005 15:34
To: dais-wg at ggf.org
Cc: Susan Malaika; Mario Antonioletti; amrey at epcc.ed.ac.uk; Norman Paton; 
Savas Parastatidis
Subject: [dais-wg] Fw: WS-DAI progress

I have previously sent the core spec out to a copy list and missed people 
out. This time I worked to get the changes done before going on leave at 
the end of May and sent it to the DAIS list and no one got it! Its getting 
worse:-) Mario thinks its to do with attaching a zip file. So I can only 
apologise but I did try and get this out to you all a week and a half ago. 
Now I am adopting the strategy of putting it up on grid forge. 

The latest core spec 
http://forge.gridforum.org/projects/dais-wg/document/Grid_Data_Service_Specification/en/8 

The latest WSDL 
http://forge.gridforum.org/projects/dais-wg/document/Grid_Data_Service_Specification_WSDL/en/1 

Simon 

Simon Laws
IBM Hursley - Emerging Technology Services

----- Forwarded by Simon Laws/UK/IBM on 08/06/2005 15:25 ----- 

Simon Laws/UK/IBM 
29/05/2005 13:10 

To
dais-wg at ggf.org 
cc

Subject
WS-DAI progress

Here is the current WS-DAI document and WSDL... 

[attachment "Grid_Data_Service_Specification-v1-27May05.doc" deleted by 
Simon Laws/UK/IBM] 
[attachment "wsdl0.7-270505.zip" deleted by Simon Laws/UK/IBM] 

Still not done but gives a good idea about how it could look. I'm out on 
leave this week so will pick up again in a week or so. 

Regards 

Simon 

Simon Laws
IBM Hursley - Emerging Technology Services
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.ogf.org/pipermail/dais-wg/attachments/20050610/8d4151c9/attachment.html 