> > On 12/20/2005 09:06:07 AM, [EMAIL PROTECTED] wrote: >> Tim Bunce writes: >> > On Mon, Dec 19, 2005 at 01:28:32PM -0500, Mladen Gogala wrote: >> > > >> > > Why is it that I can't get a response to my question: when will >> > > DBD::Oracle fully support Oracle's array interface? >> > >> > I don't get paid to work on DBD::Oracle. It's a hobby, and one that >> > generally gives me far more headaches than it could possibly be worth. >> > >> > Patches welcome. Volunteers to help maintain it also welcome. >> > Take your pick. >> > >> >> I for appreciate your work on DBI/DBD! I have been working with >> DBD::Oracle for a few years now, and I am not 100% certain what is >> being requested? What part isn't supported? > > Use of the array interface isn't supported. If I do bind_param_array and > execute_array > DBD::Oracle will break it into single step executions. Recently, I had to > load a file > of 20,000,000 records into a database. I wrote a perl script using > bind_param_array and > execute_array and it worked much slower then anticipated. The reason was > the array interface. > I had to re-write the file itself into a format that could be used by SQL > Loader and > re-do the job. Perl DBI isn't very useful as a loading tool without the > support for the > array interface. > > >> >> -- >> C Wayne Huling <[EMAIL PROTECTED]> >> >> > > -- > Mladen Gogala > http://www.mgogala.com > > Hi
I have used DBI/DBD::Oracle for many years, but I would not dream of using it to batch load 20 million rows. Oracle recommends the use of sqlldr for large batch loads, and if performance is critical use the the DIRECT option. If the data is more complicated, pre-process using a local perl script then use sqlldr, or load the data into some work tables and use a pl/sql procedures to process the data. All the best Simon
