RE: [wsdm] Terms being used in WSDM

["Sedukhin, Igor S" <Igor.Sedukhin@ca.com>]

Tom, if 1:1 is unidirectional, that is fine.

Now, a more interesting question is the relationship of WS-Resourece and "per WSDL" endpoint. Is it many to many or one endpoint to many WSRs? Theoretically it could be many to many, but I think it may be too open endded that way.

In other words, could we have same WSR exposed via several endpoints?


-- Igor Sedukhin .. (igor.sedukhin@ca.com)
-- (631) 342-4325 .. 1 CA Plaza, Islandia, NY 11788

-----Original Message-----
From: Tom Maguire [mailto:tmaguire@us.ibm.com] 
Sent: Thursday, May 27, 2004 1:32 PM
To: Sedukhin, Igor S
Cc: John DeCarlo; Wsdm (E-mail)
Subject: RE: [wsdm] Terms being used in WSDM







"Sedukhin, Igor S" <Igor.Sedukhin@ca.com> wrote on 05/27/2004 03:31:05 PM:

> Generally, +1 to your proposal, John.
>
> 4.  -> we need to add WS-Resource, not EPRs to the model. EPRs are 
> like URLs they are transient data.

+1 to adding WS-Resource as opposed to EPRs to the model.

> 4.a -> 1:1 is not possible as there are EPRs to WS-Resources that are 
> not WSDM manageable resources. For example subscription WS- Resources 
> as per WS-N are not necessarily WSDM manageable resources.

Not sure I follow this.  I don't think that the 1:1 statement was precluding the use of WS-Resource (EPRs) for things other then manageable resources but rather saying there is a 1-1 coorespondence of a manageable resource to a WS-Resource.

> I could draft changes that the introduction of WS-Resources will incur 
> in our concept diagrams.
>
> -- Igor Sedukhin .. (igor.sedukhin@ca.com)
> -- (631) 342-4325 .. 1 CA Plaza, Islandia, NY 11788
>
> -----Original Message-----
> From: John DeCarlo [mailto:jdecarlo@mitre.org]
> Sent: Thursday, May 27, 2004 11:43 AM
> To: Wsdm (E-mail)
> Subject: [wsdm] Terms being used in WSDM
>
> Hello,
>
> As was made painfully clear in today's call, the WSDM TC is using the 
> term "endpoint" in at least two to four different ways.  And this is 
> causing differences in perception of the Logical Model of MUWS.
>
> IMNSHO, this confusion is a "bad thing".
>
> While I am not a good diagram creator or modifier, I believe we need 
> to update the MUWS Logical Model to be clearer.
>
> Here is one proposal that will need to be shredded and reassembled 
> before we can get agreement, but it is a place to start.
>
> In particular, I think that William and Heather would argue that there 
> is no need any longer for the "manageability endpoint" in the WSDL 
> sense.  They might argue that we only, perhaps, talk about the "thing 
> pointed to by the Manageability EPR that accepts SOAP messages for 
> exactly one Manageable Resource".  I can't say I understand the 
> purpose, and personally find it too confusing.
> Though it does simplify in the sense that you are always only talking 
> to Manageable Resources, never to Manageability Providers (or their 
> WSDL endpoints).
>
> Anyway, here is my proposal.
>
> PROPOSAL - When we use the term "endpoint", we only use it in the WSDL 
> sense.  In all the WSDM specifications, including MUWS and MOWS.
>
> 1.  This means that we can leave all our diagrams the same, except 
> that we have to add some things like EPRs or WS-Resources.
>
> 2.  When we use the term "EPR", we say it is a reference to a 
> Manageable (Fred has a good point here that Managed is more correct 
> once you get to sending SOAP messages) Resource.  And the content of 
> the EPR is defined in WS Addressing, maybe clarified in WS-RF.
>
> 3.  Logically, to do something like GetResourceId, a Manageability 
> Consumer sends a properly formatted WSDM SOAP message, which contains 
> an EPR so that the Manageability Provider knows which Manageable 
> Resource is being referred to, to the Manageability Endpoint specified 
> in the WSDL.
>    3.a.  To remind people, this does not constrain the locus of 
> implementation at all.
>    3.b.  The Manageability Endpoint (being a WSDL Endpoint) may have 
> more than one Manageable Resource behind it.  The EPR helps out here 
> one way or another according to best practices at the time.
>    3.c.  This begs the question of the singleton pattern.  Does it 
> require an EPR or not?  And if we support the *not* case, (which helps 
> implementors that haven't gotten around to WS-RF, WS Addressing,
EPRs,
> etc.) what are the implications?  [Note:  wiser heads than I have 
> started this singleton discussion already.]
>
> 4.  The MUWS Concept Model should include EPRs.
>
>    4.a. One approach is to simply say there is a 1:1 mapping from 
> Manageable Resource to EPR.  Then you have to mention that one "thing"
> being managed may have multiple Manageable Resources/EPRs.  This is 
> what the current MUWS Concept model shows.  So we could add the EPR to 
> the Concept Model as well.  In fact, doesn't the EPR allow the 
> Manageability
>   Endpoint in the Concept model to provide access to exactly one 
> Manageable Resource?
>
>    4.b.  I don't think I like any other approaches, but will leave 
> this in here for a place holder.
>
>
> 5.  The MUWS Logical Model should address EPRs.
>
>    5.a.  One option is to say that the Manageability Consumer 
> "accesses (and provides an EPR)" the Manageability Endpoint which 
> "provides access to the Manageable Resource indicated by the EPR".
>
> --
>

T o m   M a g u i r e

STSM, On Demand Architecture
Poughkeepsie, NY  12601

internet:                 tmaguire@us.ibm.com
phone:                     845.433.9401 (t/l 293-9401)