In article <BXMcd.3646$YM4.761411_at_news4.srv.hcvlny.cv.net>,
johnp_at_optonline.net says...
> I need to write an ado app that reads a 27million record table, process it
> and insert it into a new table.
> Unfortunately I have no choice on the use of ADO.
>
> My question is, on the inserts am I better off using a Command object and
> specify not to return any record sets. Or should I use the Recordset object
> and specify lockbatchoptimistic?
>
> I know I should check to see if the 0040 driver/interface is being used.
>
> Also, for the reads and writes should I use client side or server side
> cursors. I really do not need any transaction processing. The table will not
> be accessed during the process.
>
> So far it seems that writing the data to a text file and then using the bulk
> insert utility( or whatever it is called) is the fastest but again that is
> not desired by others.
>
>
> So any feed back is greatly appreciated
>
>
Whether you use a Command or a Recordset, LockBatchOptimistic or
LockReadOnly, if you use ADO you are looking at days, not hours, to
complete this. Please consider using SQL*Loader and stored procedures to
get it down to the hours/minutes range. If you can't, just use the ADO
default settings. It won't matter anyway.
Received on Mon Oct 18 2004 - 06:56:33 CDT
