breaking some basic SOA concepts...
Quote:
Originally Posted by gmatthews
sys_B_get_info(list_of_500_ids);
or better yet.
sys_B_get_info(some_criteria_to_identify_500_objec ts);
Is it a good idea to send raw ids to the other systems?
Is it even a better idea to send some criteria? What will the other side do with these criteria? Will they execute a query on your schema? Doesn’t that violate the autonomous nature of sys_A?
Simply put - there is no way to do this in a SOA world without using some EII /Data Integration tool (Websphere Information Integration for example).
You end up using complex tools to do some basic things
Hibernate/ejb,Spring,Glue (Web Service)
I would suggest to make the application as a loosely coupled between the tiers. If the system is not going to be a data ,transaction critical system , I would suggest you guys to go for Hibernate in the DATA ACCESS LAYER. In case if you have a financial module use EJB for that specific module. Not for everything.
Web Access:
Spring gives you various advantage to access the DAO / Hibernate layer and it is easy for us to publish the object as a webservices using GLUE.
So go for HIBERNATE/EJB,SPRING and GLUE initially.
When the SOA is getting matured and proven it will be very easy for us to port the application as a SOA.
Use Webservice architecture locally within the application.On top of that we can build SOA easily.
THINK REUSABILITY WHEN YOU DESIGN.
Cheers.
Ilango Djearamane