Hi!
I'm just playing around.
We have a software which currently uses the SQL Server from MS to store data.
In the future we like to support other databases, too.
One of the other dbs will be Oracle, of course.
I use following DBs: MSDE 2000 SP4 and Oracle XE
I wrote a schema for both dbs and generated the LLBLGen Pro projects.
Since our software relies on Guids as PKs, I wrote an TypeConverter to map
the Guid to the used db type.
I tried following types in the Oracle db as Guid:
- raw
- nvarchar2(36) (I used this in the Sql Server, too)
- char(32) (I used this in the Sql Server, too)
Raw was the worst solution. I inserted 100.000 rows in both dbs
and executed following statement on both dbs:
delete from element
Oracle needed for this small script nearly 6 minutes(!) to execute whereas the
SQL server needed just a few seconds.
Representing the Guid as nvarchar2 or char cut down the time but Oracle is far slower
than the Sql Server.
So, my question is, how could I speed up the access to the Oracle db?
Has someone run into the some problems or discovered the some thing?
Curious
Christoph