i'm working on a "read only" project, we'll bench both full jdbc and hibernate prototypes.
The goal is to retrieve a lot of data and "give it" to a web service.
JDBC proto:
+ optimized queries given by BDA
+ give exactly what is needed
- if tomorrow they ask us to add other table fields.... ughhh have to study again the 50 lines query
- have to know about caching since the data are updated 2 times per day
Hibernate proto:
+ short time of development
+ pluggable cache, and with it incredible performances if the objects have been loaded...
+ you can easily update graph of object definition (so the future needs)
+/- you can optimized sql generated by knowing all about HQL (not always easy but forum helps a lot)
+/- you can reuse the optimized dba query by using SQLQuery API, but in this case.... "why using hibernate?"
- you may need to map non business objets, i'm thinking about parameter tables, for example many-to-one doesn't allow a "where attribute" which could filter an association on a parameter which is only updated every 2 years, so you have to map it. But i'm looking at the formula option... it could solve this need
If this can help you....
Anthony
|