[kepler-dev] RE: [seek-dev] EcoGrid Query Service: Instances and Factories
ludaesch at sdsc.edu
Tue Jan 20 19:41:58 PST 2004
Matt and others:
To throw one more twist into the mix: This email from Karan from the
"Grid front" is very interesting!
In the light of these events we should be careful about how we spend
our time in the EcoGrid. Maybe we should go back to something as
simple as possible that works. Then "sit out" the "standards wars"
and just have a simple, unified web service-based access to
registered EcoGrid resouces.
>>>>> "kb" == karan bhatia <karan at sdsc.edu> writes:
kb> just FYI,
kb> at globus world today, the globus alliance announced that OGSI (open
kb> grid services architecture) will be replaced with WSRF (web services
kb> resource framework). see links below for more details...
kb> - press announcement:
kb> - wsrf homepage: http://www.globus.org/wsrf/
kb> - FAQ: http://www.globus.org/wsrf/faq.asp
kb> Initially this looks like an evolution of OGSI, same basic concepts,
kb> but some important differences. Globus Toolkit 4.0 will include the
kb> wsrf implementation and is due out in third quarter of 2004. It is
kb> interesting to note that wsrf was not produced by the GGF and the ogsi
kb> working group, it was done by ibm and globus outside of any standards
kb> - karan
>>>>> "BZ" == Bing Zhu <bzhu at sdsc.edu> writes:
BZ> In today's SEEK meeting at SDSC, we also talked about the possibility for
BZ> developers other than
BZ> Ecogrid developers to build their client(s) based on WSDLs of ecogrid
BZ> services. Using of Factory
BZ> model apparently makes it impossible for others to do so.
BZ> If we have such a request from other groups, I agree with your suggestion
BZ> that we have to go
BZ> back to a non-Factory approach. Question is if this approach can handle
BZ> multiple connections
BZ> -----Original Message-----
BZ> From: seek-dev-admin at ecoinformatics.org
BZ> [mailto:seek-dev-admin at ecoinformatics.org]On Behalf Of Matt Jones
BZ> Sent: Tuesday, January 20, 2004 5:10 PM
BZ> To: seek-dev
BZ> Subject: [seek-dev] EcoGrid Query Service: Instances and Factories
BZ> EcoGrid team,
BZ> I've been working on the registry issues, and have run into some issues
BZ> with the way Jing and Bing are using the Grid "Factory" pattern. As
BZ> they have implmented it, each wrapped server (like a metacat or srb)
BZ> exposes a "Factory" which has a "createInstance" method. When called, a
BZ> new endpoint is created that exposes the methods in our
BZ> EcoGridQeuryLevelOne interface spec. This endpoint is then queried by
BZ> only that one client. However, this pattern makes it very hard for us
BZ> to discover where the EcoGridQueryLevelOne services reside, because they
BZ> don't exist at the time that the client is querying the registry.
BZ> I don't see what we gain from using the factory pattern here at all
BZ> (given our underlying server systems), so I am proposing we eliminate
BZ> use of Factories and just stick to EcoGridQueryLevelOne instances. I
BZ> think that doing so would substnatially simplify the demands on clients
BZ> in the service discovery pahse while clients are querying the registry.
BZ> The details aof how I arrived at that proposal with Jing are in the
BZ> attached IRC conversation.
BZ> Please review this information and let me know if you think it is a good
BZ> idea to eliminate the use of factories. Thanks,
BZ> PS We're on a tight time schedule, so prompt feedback is appreciated.
BZ> Matt Jones jones at nceas.ucsb.edu
BZ> http://www.nceas.ucsb.edu/ Fax: 425-920-2439 Ph: 907-789-0496
BZ> National Center for Ecological Analysis and Synthesis (NCEAS)
BZ> University of California Santa Barbara
BZ> Interested in ecological informatics? http://www.ecoinformatics.org
BZ> seek-dev mailing list
BZ> seek-dev at ecoinformatics.org
More information about the Kepler-dev