[Cryptography] Simpler programs?

[Jerry Leichter]

On Apr 16, 2014, at 7:28 PM, Lodewijk andré de la porte <l at odewijk.nl> wrote:
> For me it's about bottlenecking. Going through a small and secure place, and do all the ugly stuff there.
> 
> I'm still waiting for the waterproof dead-simple VM with simple message passing that isolates programs from one another properly.
This was one of the original goals for VM's.  VM/370 virtual machines talked to each other through fake networks and had isolated "mini-disks".

The original view of virtualization was also the basis of an A2-secure VMS implementation at DEC many years ago.  (It never shipped - the VAX died before it gained sufficient momentum, and the port to Alpha would have been a major effort.  The effort's been written up - I don't have a handy reference.)

Unfortunately, all recent VM work has gone in an entirely different direction.  Now integration between host and hypervisor is the name of the game.  Simplicity was lost long ago in favor of performance, manageability - and all sorts of extra features.

Exactly what a VM built with security as its first goal - but usability for an interesting set of cases, given modern OS's, programming styles, interaction styles, etc. - should look like would be a nice little research project.

                                                        -- Jerry