The genesis of the project

[maeste]

Hi folks,

I'm a jboss contributor (in JBossWS project) and I'd like to become an overlord contributor too. In fact I'm discussing about the project with Mark Little before Overlord was called Overlord :)

I report here my last mail to Mark summarizing some ideas we got in an irc brain storming good to share with all community (project's channel haven't a lot of users...let me say it have just 2 user :D)

Here is my mail, cut down in uninteresting part. Please, Mark, complete or comment whatever need to be clarified

[CUT]
I loved your idea to choose a couple of things to make them (hopefully) the genesis of the project.
[CUT]
I would distinguish in this email two different goals: tools for developers and the runtime governance software.

For the first one, probably the killer application and the most interesting development would be the service modeller. Definitely interesting for me, even if my current skills are stronger in other area.

Regarding the runtime governance software, I think we need a very good design and a slick web interface to present to our users. As I told you in some previous email, I'm convinced it's important to give user a unique and easy to use console. IOW it's our car's body...
I'm convinced that consoles and user interfaces are as good as good are what they have under the hood, and so it's important to have a good design of the architecture behind in this phase.
IMHO we have 3 central component in runtime governance architecture that have to cooperate and be basis for the platform.

* WS-P and its extensions for SLAs: As said I already made an
implementation for JBossWS and you are one of the author of the specification . We could be a good
team on this stuff :P.
* Repository and query language. IMHO one of the most important. It
can't be considered unrelated to UDDIv3 I think. And probably it
(or UDDI) have to be aware of our WS-P extensions.
[Mark has answered me: We need to support UDDIv3, but we also need to make sure we're not tied to it. ]
* SAM. Maybe the killer component for runtime governance if
implemented with some new approach. I recall I read something
about using semantic databases and ontology to represent and store
activity during my master thesis reading...I have to get back on
that idea, but I think RDF can have chance in this are if I well
understood what you have in mind speaking about SAM. RDF could be
a different approach than jcr also for repository...but I have to
think better about...I'm not sure it's a great idea :)
[Mark has answered me:Yes, I'm not sure about RDF for the repository either ;) ]
Well, I'd like to work on all this aspect, also because I think they have to be thought at some level together.

[And Mark answered me:
Next step: should we take the discussions to the Overlord forums? It might help get others involved too. ]

Done! now it's your time to give some feedback and be involved in the discussion ;)
Waiting for comments I'm going to take a look to DNA and lend an hand in code, since I agree with Mark saying the repository is a good starting point for my contribution.

Of course these are just a couple of ideas on which work and start discussions and thoughts. I don't know how many of these are already developed or at least thought by some others, but I think it's place where eventually share these infos.
Thanks in advance

[marklittle]

Welcome aboard Stefano :-)

[maeste]

Hi folks,
I had an interesting chat with Randall Hauch about DNA project. I summarize here some ideas about possible overlord use of DNA (and/or DNA implementations supporting overlord needs) came up on that conversation. If you don't already know some basic concept about DNA project you probably need to take a look to it to full understand some of these points:

1. Implementation of some DNA's sequencers extraction metadata information from WSDLs, UDDI, XSD, Policies and SLAs and so on.
2. Implementation of some analyser to make reports of metadatas of the previous point, for example to expose list of non respected SLAs, or failing services
3. Implementation of connector for DNA federation to collect policies, business rules and other artifact form an SCM where developer store them.
4. Implementation of a connector for DNA federation to collect runtime metadata and organize them in JCR


One of the most interesting points of DNA, IMHO not only from overlord point of view, is the federation engine which collect metadata from different sources and integrate them in a unified graph to run on it sequencers and analysers. Overlord should use this feature to aggregate infos from the SOA environment which have distribution and heterogeneity in its own nature. But of course overlord console (the server collecting runtime infos and exposing them in a web console) needs to centralize and homogenize this infos, make reports on it and eventually use it in some way. DNA federation engine and some good sequencers and analysers can do a lot of this job!

Any feed back and thought would be very appreciated. :-)

[maeste]

Transcript of chat I had with randall Hauch can be read on this post of my blog:
http://www.javalinux.it/wordpress/?p=29

[maeste]

Hi Folks,
I'm continuing to think about and investigate around runtime aspects of SOA governance.
I'm posting here to keep community informed of my progress and thought and to have, hopefully, some feedbacks.
Well I'm working mainly on the first two points of my first post here (WS-P for SLA and Repository).
About repository there is an interesting post on ESB dev forum:
http://www.jboss.com/index.html?module=bb&op=viewtopic&t=135841
As said there we (mainly for governance, but not only) need to store ESB whole message and sequence them with DNA to extract informations useful from a governance point of views, and correlate them with other infos collected and sequenced from different sources (registry, runtime sl enforcement and so on)

About SLAs, I think and analysed different solutions. WS-Policy is needed, but for sure not sufficient to achieve our goals. I think we should study and eventually adapt WS-Agreement: http://www.ogf.org/documents/GFD.107.pdf
WS-Agreement is a standard defined by Open grid forum, with grid computing in mind, but absolutely mappable on SOA, since it is sufficient generic for any kind of distributed system. I don't want to go in deep details of my thought about now, I'm preparing a document with my analysis on this standard, and I'll post it shortly, but I'd like to have feedback from you if it sounds a totally bad idea.

Any thought?

[marklittle]

I have problems with WS-Agreement on a number of fronts, not least of which is its dependency on WS-RF.

Have you looked at Policy Intents from SCA? We're doing some work on extending WS-P in OpenCSA/SCA.

[maeste]

I understand your problems with ws-rf.
What I have liked of WS-Agreement have been templates and agreement factories.

Anyway, I haven't looked yet at WS-P in OpenCSA/SCA. I'm going to read it now, is this link fine?
http://docs.oasis-open.org/opencsa/sca-policy/sca-policy-1.1.html

[marklittle]

Yes, those are fine. Though they really reference the W3C standard anyway.