Re: [Vrs-development] thoughts

[Chris Smith]

On Wednesday 27 March 2002 11:36, you wrote:
> > The cluster has to exist before agents/services can be sent
> > between LDS's.
> > The distribution of agents will be done via control messages.
> > Oh - erm - chicken and the egg.....
>
> Hmm, where's the egg?  :-)
> Yes, the cluster has to exist before you can send agents between LDSes, but
> where's the catch? Why send agents before the VRS even exists?

quote:
> > The usage of the mobile agent can be further extended
> > to handle aysnc control messages used within the
> > cluster.

Which is saying:

If you design the system so that agents act as the mechanism
by which the cluser control messages are sent..... and the cluster
control messages hold the cluster together.... but you have to
have a cluster before you can sent agents about.... but the agents
handle your cluster control messages....

So which comes first?

Basically I'm saying that the VRS architecture cannot be provided by
entities deployed into the VRS cluster.

This is not to say that we can't EXTEND the VRS functionality by
using such agents.  I think that is a very attractive idea! 
but here we're talking about using agents to provide something
fundamental to the existence of the VRS to begin with - and something
that the middleware was designed for.

Chris

-- 
Chris Smith
  Technical Architect - netFluid Technology Limited.
  "Internet Technologies, Distributed Systems and Tuxedo Consultancy"
  E: address@hidden  W: http://www.nfluid.co.uk