Subject: RE: [wsdm] What comes first: The Sandwich Approach to Development
Management Through Web Services is necessary for any result from this group, as the result must be implementable. Nevertheless, Management Through Web Services not sufficient, because without a knock-em-dead Management Of Web Services specification, others will creatively solve the management of issue, and the result of this group will be yet another standard.
Management Of Web Services is where this group can make a real contribution to the state of the art in support of other groups such as W3C, GGF, and DTMF. It provides the vision for what management through is to accomplish, without which management through is directionless. For me, the differentiator for Management of Web Services is an optional provision for initiating a back-channel flow of information from the Management Environment to the Web Service regarding the resources upon which the Web Service is dependent.
Traditionally, a way of solving impasses like this is the sandwich model of development – start at both the top and bottom and work toward the middle. With that spirit, I propose the following strawman approach:
1) A period of vision setting involving all members of the group.
2) A period of rotating top-down (Management Of Web Services), bottom-up (Management Through Web Services), and joint sessions. Some participants have expressed primary interest in one of the areas; other participants have said they need to participate in both. Alternating weeks is a way of accommodating these interests without overburdening those who wish to participate in all sessions.
3) A period of convergence.
There are great opportunities here as we draw on everyone’s strengths.
From: M. Homayoun Pourheidari [mailto:firstname.lastname@example.org]
Sent: Friday, April 11, 2003 5:45 PM
Cc: email@example.com; Richard_Nikula@bmc.com; firstname.lastname@example.org; email@example.com; firstname.lastname@example.org; Ereed@novell.com; email@example.com; MAIERA@de.ibm.com; firstname.lastname@example.org; email@example.com; firstname.lastname@example.org; email@example.com; firstname.lastname@example.org; email@example.com; Igor.Sedukhin@ca.com; firstname.lastname@example.org; email@example.com; firstname.lastname@example.org; email@example.com; firstname.lastname@example.org; email@example.com; firstname.lastname@example.org; email@example.com; firstname.lastname@example.org; Joe.Head@commerceone.com; email@example.com; firstname.lastname@example.org; email@example.com; firstname.lastname@example.org; email@example.com; firstname.lastname@example.org; Jeff.Hodges@sun.com; email@example.com; firstname.lastname@example.org; email@example.com; firstname.lastname@example.org; email@example.com; firstname.lastname@example.org; email@example.com; swhite@SeeBeyond.com; firstname.lastname@example.org; Alok.Z.Srivastava@oracle.com; email@example.com; firstname.lastname@example.org; email@example.com; Darran.Rolls@waveset.com; firstname.lastname@example.org; email@example.com; firstname.lastname@example.org; email@example.com; firstname.lastname@example.org; email@example.com; firstname.lastname@example.org
Subject: Re: [wsdm] What comes first
I'm having the same problems as Pankaj to post to the wsdm mailing list. I hope someone looks into this...
--I agree that designing a solution for mgmt. of web services is an end goal that this TC should focus on. However, I don't see how we can do a good job of that without considering how this information is retrieved, accessed, and exposed and how what we have modeled is implemented. If I can make an analogy, it is like saying that we want to build a 5 story building with all sorts of features and we will worry about the material we use to build this building later! Unfortunately, many features that we may want to build into this house won't be supported by the intrinsic behaviors and capabilities of some materials and no matter how nice the house looks in our mind, its construction will fail because we have not considered the features of the materials and the infrastructure that we may likely use to build this house. [ BTW: If you ask any good building constructor, they will give you the same argument :) There is also a great story about a large auto company that tried to build a revolutionary car without considering the material that should be used to build it. The project was a big failure... I can send it separtely to whoever that is interested]. We are also likely to over engineer somethings if we don't take some features of the building blocks and the infrastructure into account and create an unnecessarily heavy weight solution in addition to introducing other types of problems (e.g. lack of flexibility, scalability, etc.) So, I strongly believe that we will not be able to create a successful solution if we do one (mgmt. of) without having an eye on the other (mgmt. through). I also like to caution everyone to not jump on any bandwagons without considering the requirements of what we need to accomplish. We should certainly consider work that is done and being done elsewhere (e.g. W3C, DMTF, GGF, etc.) but having wrapped our minds around them is very pre-mature before we even look at their pros and cons in the context of this TC and whether they even meet our needs. Regards, H. ---
GARG Shishir / FTR&D / US wrote:
Well put, Mark. I agree that the "of" aspect is clearly the first step. As some approaches show today, using a random method that is read-only to access the Web service's availability and performance metrics are not the optimal mechanism. So the first step would be to standardize on some manageability OF these WS.
After that, these results can be readily applied to the management "by" Web services, but its a much more gray subject and hard to quantify what that might look like and how widely applicable it would be (depending on how much ocean is left :).
France Telecom R&D.
> -----Original Message-----
> From: Mark Potts [mailto:email@example.com]
> Sent: Friday, April 11, 2003 1:14 PM
> To: Nikula, Richard; firstname.lastname@example.org
> Subject: RE: [wsdm] What comes first
> The Management "of" Web services by definition and requirements
> specified within the W3C is consistent with WSA (web services
> architecture) and therefore will "use" Web services. I agree that the
> first step is a management information model specific to Web service
> resources (endpoints). Using the model defined for the management "of"
> Web services we can then codify the model (WSDL, XSD etc) such that it
> meets the defined requirements and is consistent with other
> surrounding the "use" of Web services for management (GGF, DMTF etc).
> This combined work can then be used and leveraged when
> looking at other
> types of resources below and above the services layer that
> make sense to
> be managed "using" Web services.
> There are good reasons to adopt this approach, mostly driven by
> producing something meaningful and useful to the Web service and
> Management community in a timely fashion;
> 1) To attack Management "using" Web services is a vast and abstract
> task. There will many levels of management specific to the resources
> being managed (element level through services and applications to
> logical compositions and processes) and to abstract this will take an
> incredible amount of time and effort (even leveraging past
> work in this
> area). We should work towards this but not try to boil the ocean
> initially - divide and conquer.
> 2) Managing of Web services using Web services is a natural starting
> point, we will discover a lot about what can be done within
> the confines
> of a specific resources, and the scope of the effort itself is
> manageable and achievable given our time frames.
> Mark Potts
> Talking Blocks
> > -----Original Message-----
> > From: Nikula, Richard [mailto:Richard_Nikula@bmc.com]
> > Sent: Friday, April 11, 2003 11:25 AM
> > To: email@example.com
> > Subject: [wsdm] What comes first
> > I think that a more clear definition of the deliverables is
> > required to answer which to do first. Our goal in creating
> > the new charter was to leverage/augment work in DMTF, W3C and
> > GGF in particular. We need to define what that means, then
> > based on the readiness of the groups, go from there. I think
> > a lot of our work is figuring how and if we can work within
> > the framework we propose.
> > Based on discussions in the old TC, while somewhat
> > simplistically stated, I would have thought the first thing
> > we would do is to take the work done by the W3C in terms of
> > resources and methods; and working "with" the application
> > modeling team from DMTF, produce a web services model
> > (management of). Once we had the web services model, we
> > could then proceed based on the work the GGF was doing to
> > produce a WSDL based implementation of the service
> > (management using).
> > Richard
>-- M. Homayoun Pourheidari Web Services Management Operation HP OpenView Division 408.447.5012 firstname.lastname@example.org