For a million other reasons (performance, performance, performance),
transactions should limit themselves to the same server as much as
possible. When they cannot, XA deals with that by sharing your
transaction across multiple connections (at least Oracle and Sybase
should).
But, if you are updating the same database identity in two servers, all
bets are off. And this is most likely to happen with two entity beans of
two different types, and most likely two EJB servers from two different
vendors!
(Which is why I don't believe in the distributed cache model)
>
> > Also, what happens if T1 is using a remote reference Ra, T2 is using a
> > remote reference Rb, and now T1 is using Rb as well?
My question was, if T1 uses Ra and Rb, and T2 is using Rb (we're talking
about the same identity and same bean type, just different references),
does:
T1 uses A through Ra, B through Rb, and T2 uses B through Rb, or
T1 uses A through Ra, A through Rb, and T2 uses B through Rb,
In the first case, not even the XA magic can keep you away from harm. In
the second case, you get delegation from reference to instance (a la
stateless beans), which is not clearly specified in the specs.
arkin
===========================================================================
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".
