[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [dist-obj] Why use Application Servers?
Kevin Dick wrote:
>I believe Persistenhce also has a framework for
> updating the cache in the face of outside
> database updates based on triggers and TIB/Rendezvous.
Yes they do. It is called cache synch. Check out the
INSTINET case study at:
It really blows away most other EJB implementations as
far as being able to connect multiple app server clusters
and making them work together. As far as a transaction
service I still prefer BEA's, but that's another thread.
Before we get too far off track, I'd like to question *exactly* how
outside db updates are handled (as people on this list can probably
tell, that's a subject near and dear to my heart). Last I looked,
several popular database vendors' products could hardly be considered
"active". Can someone summarize the technical bits? (Triggers are
useful, but they only get you halfway -- you still have to get from
the db process set to the app-server process set. Worse, I've never
seen a relational db. trigger implementation that had sufficient
expressive power and handled exceptions and errors gracefully -- ever
had to deal with a tablespace error in a trigger?)
Note that persistence is an unusual case. It is easy to make the
argument for such additions if you buy-into their object-relational
mapping layer, and predicate caching. But I know of at least one
fortune 500 company where their OR technology resulted in a 50Meg
library that was deemed unsuitable (the company still uses it but not
as widely as was once h(y)oped).
And trading applications are another wierd special case -- a lot of
people are surprised when they find out that "fire and forget" does
work in an amazing number of cases -- but be careful of generalizing
To manage your subscription, mailto:email@example.com
Archives, FAQ, etc. http://www.distributedcoalition.org/mailing_lists/