You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@tuscany.apache.org by "Antollini, Mario" <ma...@intel.com> on 2007/03/16 22:03:25 UTC

Understanding Service Discovery

Hello Meeraj,

 

I have read several emails and I got to know that you are working on the
Discovery service. I am very interested in this topic and I will like to
get a better understanding about it. 

 

I have seen that you have also mentioned the Federated Deployer,
Federated Assembly and Marshaling Framework throughout your posts. I
think all these components are somewhat connected to the Service
Discovery mechanism, aren't they? I will really appreciate if you can
give me some details on how they all interact, if they do so at all.
And, I will also like to understand the big picture here; i.e. what is
the main goal you are trying to achieve with all that. In addition to
that I will like to understand the role JXTA plays here and how it also
interacts with these components.

 

Finally, I would like to know which is its is (i.e.; what it is already
able to do, what parts are missing, if you are willing to receive any
help, you are planning to get it ready for M3, etc).

 

As you saw I am asking way too many things, however I will be really
glad if you can answer at least some of these questions.

 

Thanks a lot in advance!

 

Looking forward to your reply,

 

Mario E. Antollini

Intel Software

ASDC

+54 351 414 5594

mario.antollini@intel.com <ma...@intel.com> 

 

 


Re: Understanding Service Discovery

Posted by Meeraj Kunnumpurath <m....@hotmail.co.uk>.
Thanks Jeremy.

Mario, I think Jeremy has covered most of your points. If you need any more 
clarifications or have further queries, please feel free to ask. As Jeremy 
said, any help in these areas, ongoing and/or especially in the next two 
days :) will be highly appreciated.

Ta
Meeraj


>From: Jeremy Boynes <jb...@apache.org>
>Reply-To: tuscany-dev@ws.apache.org
>To: tuscany-dev@ws.apache.org
>Subject: Re: Understanding Service Discovery
>Date: Fri, 16 Mar 2007 18:53:14 -0700
>
>On Mar 16, 2007, at 2:03 PM, Antollini, Mario wrote:
>
>>Hello Meeraj,
>>
>>I have read several emails and I got to know that you are working  on the
>>Discovery service. I am very interested in this topic and I will  like to
>>get a better understanding about it.
>
>Great to have you involved.
>
>>
>>I have seen that you have also mentioned the Federated Deployer,
>>Federated Assembly and Marshaling Framework throughout your posts. I
>>think all these components are somewhat connected to the Service
>>Discovery mechanism, aren't they? I will really appreciate if you can
>>give me some details on how they all interact, if they do so at all.
>>And, I will also like to understand the big picture here; i.e. what is
>>the main goal you are trying to achieve with all that. In addition to
>>that I will like to understand the role JXTA plays here and how it  also
>>interacts with these components.
>
>Their interaction is only indirect - they are more just using it as a  
>transport.
>
>The discovery layer is responsible for determining which runtime  processes 
>are participating in the federation at any time. We're  using JXTA for that 
>as it is a standard peer-to-peer protocol and is  available in many 
>different languages, specifically both Java and C++  which are the two 
>runtime implementations we have in Tuscany. We also  want other protocols 
>to work as well (for example, Bonjour or JINI)  but want to get one working 
>first.
>
>As well as tracking the available processes it also provides a  mechanism 
>for allowing those processes to exchange messages and we  are using that to 
>send management operations across the system fabric.
>
>Some of those management operations are associated with federated  
>deployment. SCA's deployment model is based on the concept of changes  to 
>the domain assembly (e.g. include composite). Our implementation  of that 
>routes those changes to the controller nodes (at first one,  next 
>replicated, finally distributed) which take those logical  changes and 
>convert them to the physical ones that need to be made on  participating 
>runtimes. With the JXTA-based fabric, those are XML  encoded JXTA messages; 
>the marshalling framework is all about  converting between those messages 
>and the internal data structures  (physical model) used by the Java runtime 
>(C++ does not support this  yet).
>
>The federated deployer is a runtime node service that receives  
>demarshalled change set messages from the fabric and applies the  changes 
>they contain to the local runtime - basically creating,  removing 
>resources, components and wires.
>
>At the moment there's no interaction between JXTA and user components  but 
>if that was useful then we would just need to create a transport  component 
>implementation and wire attacher on the physical side and  the binding 
>support on the controller side.
>
>>
>>
>>Finally, I would like to know which is its is (i.e.; what it is  already
>>able to do, what parts are missing, if you are willing to receive any
>>help, you are planning to get it ready for M3, etc).
>
>We are in heavy development right now planning to show it working  next 
>week at TSSS - any help would be appreciated, both now and  ongoing. All of 
>this is happening in trunk for the 2.0-alpha2 release.
>
>>
>>As you saw I am asking way too many things, however I will be really
>>glad if you can answer at least some of these questions.
>
>Hope I covered the key points.
>--
>Jeremy
>
>
>---------------------------------------------------------------------
>To unsubscribe, e-mail: tuscany-dev-unsubscribe@ws.apache.org
>For additional commands, e-mail: tuscany-dev-help@ws.apache.org
>

_________________________________________________________________
MSN Hotmail is evolving - check out the new Windows Live Mail.  
http://ideas.live.co.uk/


---------------------------------------------------------------------
To unsubscribe, e-mail: tuscany-dev-unsubscribe@ws.apache.org
For additional commands, e-mail: tuscany-dev-help@ws.apache.org


Re: Understanding Service Discovery

Posted by Jeremy Boynes <jb...@apache.org>.
On Mar 16, 2007, at 2:03 PM, Antollini, Mario wrote:

> Hello Meeraj,
>
> I have read several emails and I got to know that you are working  
> on the
> Discovery service. I am very interested in this topic and I will  
> like to
> get a better understanding about it.

Great to have you involved.

>
> I have seen that you have also mentioned the Federated Deployer,
> Federated Assembly and Marshaling Framework throughout your posts. I
> think all these components are somewhat connected to the Service
> Discovery mechanism, aren't they? I will really appreciate if you can
> give me some details on how they all interact, if they do so at all.
> And, I will also like to understand the big picture here; i.e. what is
> the main goal you are trying to achieve with all that. In addition to
> that I will like to understand the role JXTA plays here and how it  
> also
> interacts with these components.

Their interaction is only indirect - they are more just using it as a  
transport.

The discovery layer is responsible for determining which runtime  
processes are participating in the federation at any time. We're  
using JXTA for that as it is a standard peer-to-peer protocol and is  
available in many different languages, specifically both Java and C++  
which are the two runtime implementations we have in Tuscany. We also  
want other protocols to work as well (for example, Bonjour or JINI)  
but want to get one working first.

As well as tracking the available processes it also provides a  
mechanism for allowing those processes to exchange messages and we  
are using that to send management operations across the system fabric.

Some of those management operations are associated with federated  
deployment. SCA's deployment model is based on the concept of changes  
to the domain assembly (e.g. include composite). Our implementation  
of that routes those changes to the controller nodes (at first one,  
next replicated, finally distributed) which take those logical  
changes and convert them to the physical ones that need to be made on  
participating runtimes. With the JXTA-based fabric, those are XML  
encoded JXTA messages; the marshalling framework is all about  
converting between those messages and the internal data structures  
(physical model) used by the Java runtime (C++ does not support this  
yet).

The federated deployer is a runtime node service that receives  
demarshalled change set messages from the fabric and applies the  
changes they contain to the local runtime - basically creating,  
removing resources, components and wires.

At the moment there's no interaction between JXTA and user components  
but if that was useful then we would just need to create a transport  
component implementation and wire attacher on the physical side and  
the binding support on the controller side.

>
>
> Finally, I would like to know which is its is (i.e.; what it is  
> already
> able to do, what parts are missing, if you are willing to receive any
> help, you are planning to get it ready for M3, etc).

We are in heavy development right now planning to show it working  
next week at TSSS - any help would be appreciated, both now and  
ongoing. All of this is happening in trunk for the 2.0-alpha2 release.

>
> As you saw I am asking way too many things, however I will be really
> glad if you can answer at least some of these questions.

Hope I covered the key points.
--
Jeremy


---------------------------------------------------------------------
To unsubscribe, e-mail: tuscany-dev-unsubscribe@ws.apache.org
For additional commands, e-mail: tuscany-dev-help@ws.apache.org