I am looking to implement a system that will take a flat file of table rows
and apply them to a database. The file will know the table name and the 1-3
primary keys and will be in primary key order.
The old system worked on the premise
- It sucked out the database into a structure in memory
- Compared keys of the flat file to that of the memory image and did one of
two things.
- if not found deleted the record in the database.
- compared columns and updated appropriately
- It would then reread the flat file and insert all records not present in
the image file.
This process needs to be very very efficient so my question are:
- Under 7.3.4 and 8i does it actually perform an update if values are the
same for all column. This would remove the requirement to compare indivual
columns
- Can I read from a 7.3.4 database in primary key order using cursors. Some
tables have over a million records and sorting is hell.
(I ask this because I seem to remember that I was forced to add a where
clause to force an index to be used once).
Any design examples welcome.
Regards,
Iain
Received on Wed Nov 14 2001 - 01:17:06 CST