CMR with optimal caching

[alwyn]

Hi,

I'm trying to implement a 1 to * CMR between 2 beans making optimal use of caching.

The system is used to map specific SMS keywords to specific session beans and the parameters to certain variables.

Bean A fields:
Keyword (keyword coming from phone),
ServiceCode (application identifier),
Handler (Jndi of session-bean that will handle keyword)

According to my understanding of the caching, the best way is to not use a finder but use getByPrimaryKey. So the PK for Bean A would be a composite of Keyword and ServiceCode so that the database do not need
to be queried for a PK to find the bean in the cache with.

Bean B maps the keyword parameters.
Bean B fields:
offset (offset from left to right from the keyword),
parameter (variable to map it to)

Now there is obviously something missing, the foreign key to Bean A and perhaps even a PK for bean B.

What needs to be added to Bean B to get this working? It is important that the solution would optimize cache and database access too.

Best regards,

[triathlon98]

For starters, you have to use commit option A (or possibly D depending on your app). B/C does not cache (as the data may have changed in the DB).

Joachim

[alwyn]

I am using commit option D. Most of the problem is finding and using the correct primary keys to bypass doing a SELECT on all keys which is what a finder would do.