[cap-talk] How desirable / feasible is a persistent OCAP language?

James A. Donald jamesd at echeque.com
Fri Jul 18 21:18:27 CDT 2008


Ivan Krstić wrote:
> On Jul 17, 2008, at 10:25 PM, James A. Donald wrote:
>> If any state needs to be consistent and persistent, that
>> state can only be in one place.  If one finds one has
>> distributed state that needs to be consistent between
>> nodes, one is in deep trouble.  Sooner or later, to
>> maintain consistency, will have to end persistence,
>> shutting down the whole shebang and starting over.
> 
> Indeed. Pat Helland's paper on the matter is well worth a read:
> 
>      <http://www-db.cs.wisc.edu/cidr/cidr2007/papers/cidr07p15.pdf>


Thanks.

The world is just starting to grapple with the problem of massively 
networked computing power.

Knuth taught us what to do when 256 bytes was no longer a large amount 
of memory.  We still are not sure what to do when 256 nodes is no longer 
a large network.


More information about the cap-talk mailing list