| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
Home -> Community -> Usenet -> comp.databases.theory -> Re: S.O.D.A. database Query API - call for comments
Robert Sundström wrote:
> But, I have read some of the code of your benchmark and noticed that you
have
> chosen to dynamically build _each_ SQL-statement and use JDBC-Statement
> objects for executing them. This is (just as you say) expensive. I would
guess
> that a typical SQL-database will spend 90% of its time compiling and
> generating query plans for your statements.
Sorry for the simplicity of the benchmark. I know that the SQL part can be optimized for the specific task that is taking place. The 1000 runs try to simulate insert, update and retrieval of 1000 *different* objects.
> The idea of SQL to provide a generic, dynamic and easy to understand way
of
> manipulating databases. SQL-statements *can* be compiled and executed
> dynamically, just as you do in your benchmark, but the preferred way is to
> compile the statements in advance using parameter markers and execute them
> multiple times.
O.K. we could add a prepared-statement version also, and compare performance.
On the other hand our engine also has a great disadvantage in the simple
benchmark:
It stores an integer/String *plus* an object to carry it.
> Compare this to your object oriented database approach. When you have
compiled
> your java database program, all possible database operations are set
> and defined at compile time. You will never dynamically construct and
compile
> Java code!
No, we do not dynamically change Java code.
A side question:
Do OR-mappers typically use dynamic statements or prepared statements?
Kind regards,
Carl
--- Carl Rosenberger db4o - database for objects - http://www.db4o.comReceived on Fri May 04 2001 - 09:23:40 CDT
![]() |
![]() |