Re: [Vrs-development] thoughts

[Tim Terlegård]

> Assuming that the administrator has privilages on both
> machines, how does this differ from simply executing a
> resident monitoring program on the observed machine
> that does exactly what the agent would do.  Then it
> phones home with the answers?

I knew you were going to say this  :-)

1. I might want to send the MA to a server I can't admin. Let's say I'm a 
network manager and monitors the network of an organisation. I could do all 
the monitoring with MAs, removing the need for root-privileges.
2. Sitting on a train, downloading bytecode (to the mobile phone) for a game 
and running it. I'm not sure if that is an MA, but could be.
3. I want some task done. I program a little MA and send it, then I go 
offline. The next day when I arriva at work the MA notices I come online and 
returns the results.
4. An MA can collect data from several servers (without going back home), 
filter out unnecessary information and compute the results. Only the results 
will go back to the sender and the network load decreases.

You said that VRS will be able to execute DataSets and move them around the 
net. I guess this is something like like Mobile Agents. I agree, though, that 
Mobile Agents aren't absolutely necessary.


> > Well, I don't know what value you want   :-)
> > What comes to my mind (because of the similar
> > project goals) is that VRS
> > could be a platform supporting p2p, mobile agents,
> > vpn, fault tolerance,
> > replication, distribution of power and space. SEE
> > builds (whatever they want)
> > on top of this, because everything you possibly
> > could want is right there.
> > This I'd like, but it may not be what you guys want
> > or what DotGNU wants.
>
> That doesn't sound inconsistant at all with what we
> are thinking about.  Do you see any basic archectural
> problems in the docs so far in designing all of this?

Nopes, not at all. If all aspects of security is integrated into VRS, then 
it'd be very similar.

-- Tim