OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help

soa-rm message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]

Subject: Re: [soa-rm] Requesters vs. Consumers

I would argue that clarity on the owning relationship is critical to 
large scale systems. Consider an ASP that is hosting the ERP systems of 
many businesses -- using shared resources to achieve economies of 
scale. A given computer/software component may be acting on behalf of 
any number of owners in a dynamically changing fashion. Currently, this 
is sorted out by not sharing -- by using different computers and/or 
different software invocations.

Again, consider the situation of a third party service: delivery 
services or brokers for example. Such a service is acting on behalf of 
many players, and may have a quite separate owner. At different times 
the same software/hardware entity is acting on behalf of different 
entities (consumers or providers).

One simple definition of ownership revolves around control: control is 
a relationship between an agent and a resource that gives the *right* 
to manage and use some aspect of a resource. Ownership is a 
transferrable right: you have the right to control the resource, and 
the right to transfer that right. Without either control or 
transferability you don't have ownership. (Renting gives the right to 
control, but usually not the right to transfer that right)

This stuff is definitely at the edges of our concern here. But, I 
believe that explicit, machine processable, policies are part of the 
future -- if only to help us manage large systems.


On Apr 1, 2005, at 9:13 AM, Metz Rebekah wrote:

> Frank -
> I think the concept of an un-owned agent points at the basis
> relationship between an agent and the 'owner' as "claiming ownership
> of".   The other potential relationship could be one of 'identifiable
> ownership.'  That is the owner does in fact exist, but no means to
> identify who that owner is can be established, perhaps for privacy
> purposes.
> Another question that seems to be cropping up in various forms is
> centered on the position/importance/need (can't quite select the right
> word in there) for characterizing the relationship (and thus
> distinction) between an agent and owner.
> Rebekah
> Rebekah Metz
> Associate
> Booz Allen Hamilton
> Voice:  (703) 377-1471
> Fax:     (703) 902-3457
>> -----Original Message-----
>> From: Francis McCabe [mailto:fgm@fla.fujitsu.com]
>> Sent: Friday, April 01, 2005 12:04 PM
>> To: Thomas Erl
>> Cc: soa-rm@lists.oasis-open.org
>> Subject: Re: [soa-rm] Requesters vs. Consumers
>> We originally wanted to use the term *legal entity* to represent the
>> 'owner' of the agent(s) participating. However, we were advised by
>> W3C's legal whatever that this was not a good choice. (Too politically
>> charged apparently); there was also the possibility of an un-owned
>> agent participating (the mind boggles a bit at this). However, in
>> common usage, legal entity includes people and corporations.
>> This is a tricky area, on the one hand it seems blinkered to pretend
>> that we are not designing systems for and on behalf of people. On the
>> other hand, taking people fully into account seems to take us into
>> realms where our expertise is not appropriate.
>> Frank
>> On Mar 31, 2005, at 5:17 PM, Thomas Erl wrote:
>>> It's probably a good time to think about which term we should use to
>>> represent the potential element responsible for invoking or
> initiating
>>> a
>>> conversation with a service acting as the service provider.
> Regardless
>>> of
>>> whether this becomes an "official" element within our reference
> model,
>>> we
>>> will likely need to reference such an element in our documentation.
>>> Below are some considerations we can take into account:
>>> - Both of the position papers submitted so far incorporate the term
>>> "consumer". This term is also used in the ebSOA specification.
>>> - The W3C Web Services Architecture document submitted by Frank
> McCabe
>>> uses
>>> the term "requester" and further qualifies it by suffixing it with
>>> "entity"
>>> or "agent" to represent the owner and software program respectively.
>>> (Prior
>>> to the current version of the W3C Working Note, this document used
> the
>>> term
>>> "service requester" instead of "requester agent".)
>>> - The W3C Web Services Glossary does not provide a definition for
>>> "consumer",
>>> but defines "requester agent" as follows: "A software agent that
>>> wishes to
>>> interact with a provider agent in order to request that a task be
>>> performed
>>> on behalf of its owner - the requester entity."
>>> - The term "requester agent" is used in the W3C WSDL 2.0
> specification,
>>>  whereas "consumer" is used in the WSDL 1.1 version.
>>> - The definitions document submitted by Rebekah uses the term
>>> "requester",
>>> most likely because the initial set of definitions were provided by
>>> Frank.
>>> Given that we are seeking industry-wide acceptance of our reference
>>> model,
>>> there may be a benefit to keeping our terminology in alignment with
>>> terms
>>> already in use by established (albeit implementation-specific)
>>> specifications. I personally have no preference, but I do recommend
> we
>>> decide on one term and then consider adding a definition to our
>>> glossary. We
>>> may want to leverage some of the work performed by the W3C Working
>>> Group and
>>> decide whether we also need separate terms to distinguish owner from
>>> implementation.
>>> On a related note, we have not yet discussed the concept of a
> service
>>> or
>>> service agent assuming provider and requester/consumer roles. Such a
>>> concept
>>> would also affect our definitions.
>>> Thomas

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]