Assaf,
I think we agree on everything excepting:
> -----Original Message-----
> snip...
> Once you made that assumption, the only two models I see working are
> those that I described, of hard locking and dirty checking. Hard locking
> does not require a shared cache, since you always synchronize with the
> database (just in case). Dirty checking does not require a shared cache,
> since it can always be invalid to begin with.
>
> Considering the cost (object passing over TCP/IP) I just fail to see the
> benefit.
>
The benefit is that keeping the remote caches in synch lowers the potential
for dirty check failures. Presumably it is better to have an updated cache
if possible. And yes this is offset by the cost of keeping the caches in
synch in the first case. Depends on update frequency...
As a side note we don't broadcast changes at the object level, but rather
synch at the page level. Makes for fewer network round trips which is where
you usually get burned.
-Chris.
===========================================================================
To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
of the message "signoff EJB-INTEREST". For general help, send email to
[EMAIL PROTECTED] and include in the body of the message "help".
