Re: delete five BIG tables and insert new records on same tables

Jay wrote:
> Thanks, Mark,
>
> I'm not so sure how bulk binds can work with CSV file.
> and I said i want to minimize downtime ONLY IF I have to have down time.
>
> -Jay
>
>

[...]

"down time" is a relative term. Only you know how these tables are used

• continuous access 24 x 7 for OLTP, once a month for historical reporting, all rows accessed or only the most recent ones, relational integrity constraints that cannot be disabled, indexes that must exist at all times for acceptable performance, whatever. "down time" will depend specifically on these and most likely other factors too.

You've now received several suggestions from myself and other posters, you're going to have to experiment or otherwise intuit which one best meets your needs -- or provide more detail on the specific nature of the data and how it is used in your application.

The suggestion for a bulk bind was meant to address the speed issue, as an alternate to inserting one row per SQL statement as in your original example. You would read many rows into memory (from a CSV file or whatever your ultimate data source is, you didn't explain) and then stuff them into table all at once (if I've mis-stated how the bulk bind mechanism works, I'm sure someone will correct me). Although now that I think about it, you said version 8i and the features I'm referring to I think are only available in 9i.

-Mark Bole Received on Mon Nov 15 2004 - 15:29:32 CST