[e-lang] VatTP and Waterken protocols compared

[Mark Miller]

On Fri, Sep 19, 2008 at 5:54 PM, Bill Frantz <frantz at pwpconsult.com> wrote:

>
> Both systems only save persistent state between vat "turns". It is
> not clear from the (non)documentation if the vat queue is included
> in what is saved by the persistence mechanism. (And I can't
> remember that detail.)


It is not, which is a consistent reflection of the difference in persistence
philosophy between the systems. Waterken provides local orthogonal
persistence, and distributed orthogonal-as-possible persistence. E's
persistence is non-orthogonal, either locally or distributed. Rather,
persistent E applications are responsible for re-establishing a consistent
state on revival from checkpoint. Distributed E application are responsible
for re-establishing consistent distributed state when recovering from
partition. Persistent distributed E applications must do both on revival
from checkpoint, since a crash-and-revival necessarily implies partition.

The pros and cons of the differences in philosophy are complex, and it is by
no means clear when one should prefer which. Having programmed some in each,
I was surprised by how much simpler Waterken's stance turned out to be when
it worked -- which was quite often.

The acid test of the two philosophies will be their friendliness to upgrade.
On this, the jury is still out. (Research opportunity!)

-- 
Text by me above is hereby placed in the public domain

Cheers,
--MarkM
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.eros-os.org/pipermail/e-lang/attachments/20080919/53c786b8/attachment.html