Hosted-code initiated/encapsulated distribution?

Erik Klintskog erik at sics.se
Wed Jan 21 11:56:01 CET 2004


Konstantin Popov wrote:

> Dear Nathaniel,
>
> > Salutations,
> >       If my understanding of Mozart/Oz is correct, when the VM gets a
> > ticket for a store entity on a remote host, the VM itself will initaite
> > the TCP (invariably TCP, yes?) connection to the remote host.
> >
> >       Would it be possible (and would it be smart) to allow the Oz code
> > itself to register a filehandle/socket/callback function with the VM so
> > that tickets of a given type will cause this Oz code to carry out the
> > connection request?
> >
> >       It would be trivial, I think, to reimplement the TCP-based current
> > method in Oz if this mechanism were available, as well as more general
> > mechanisms that could offer security or other guarantees.  For example,
> > two VMs could establish a TLS-secured connection and conduct all
> > distribution across that.
> >
> >       To this end, the ticket format would have to be expanded to allow
> > VMs to have non-IP based addresses, the VM would have to have a "routing
> > table" of sorts ("tickets for VM-0xdeadbeef are routed over ..."), and the
> > hosted code would have to be able to get at this identity in order to
> > initiate and set up these tables.
> >
> >       Thoughts?
> a part of this is already implemented and used: the connection
> establishment procedures (for both the "initiator" and "acceptor"
> sites) are coded in Oz. Once a connection (the standard medium are
> sockets) is established, it is "handed over" to the distribution
> subsystem.
>

This functionality/service is even documented, see:

http://www.mozart-oz.org/documentation/ds_white_paper/index.html

In this document you will find a section that describes how the ticket
mechanism
is realised, using distributed ports. Furthermore, the concept of
ConnectionProcedures
and AcceptProcedures is described.

/Erik


>
> Currently there is only a single fixed type of tickets. Given the just
> mentioned mechanism, that is quite easy to change. I'd say we were not
> really urged to offer that flexibility.
>
> Cheers,
>
>  --- Kostja.
> -
> Please send submissions to hackers at mozart-oz.org
> and administriva mail to hackers-request at mozart-oz.org.
> The Mozart Oz web site is at http://www.mozart-oz.org/.
-
Please send submissions to hackers at mozart-oz.org
and administriva mail to hackers-request at mozart-oz.org.
The Mozart Oz web site is at http://www.mozart-oz.org/.





More information about the mozart-hackers mailing list