Jason van Zyl <[EMAIL PROTECTED]> writes:

> On 12/14/01 11:24 AM, "Chris Shenton" <[EMAIL PROTECTED]> wrote:
>
>> Seems the caching should be done by the DB machine so the views it
>> presents are consistent to front-end boxes.

That would be re-inventing EJBs, and IMHO is not the way to go.  Any
RDBMS/OS implementation worth its salt can do this caching to some
extent anyhow.  The main expense is the object
construction/destruction (all that pushing, popping, and processing
gives both the heap and your CPU a work out), and for that you need in
memory caching.

> I think it can be done in a distributed fashion, but Aaron has a lot more
> experience in this area than I so maybe he can answer your question with
> some degree of authority.

For SourceCast, we've found that it makes the most sense to maintain a
LRU-based cache on each application server in the cluster, and perform
cache invalidation across the network (preferably using a reliable
multicast based implementation of a JMS or JMS-act-alike system).
This allows each app server to have an up to date cache of business
objects, and optmizes for the common case (read-only access to your
data and business rules).

--
To unsubscribe, e-mail:   <mailto:[EMAIL PROTECTED]>
For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>

Reply via email to