| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
|  |  | |||
Home -> Community -> Mailing Lists -> Oracle-L -> Re: What is the fastest way to dump oracle data into a human
Options #1 (Perl) and #2 (PRO*C) would be fastest and easiest.  The PRO*C
demo programs provide a decent start, for option #2.  Option #3 (OCI) would
be not faster than PRO*C and, due to the increased complexity of OCI, a more
problematic approach.
SQL*Plus is the easiest method to implement by far, but the SQL*Plus program introduces a surprising amount of "overhead" processing, making it much slower than Perl or PRO*C, but much faster than UTL_FILE. If you are not in a big time crunch but want decent performance, use SQL*Plus. If you are in a time crunch and need the best performance, use Perl or PRO*C.
on 1/10/04 6:09 PM, Guang Mei at gmei_at_incyte.com wrote:
> Hi:
> 
> I have a program (running on oracle 8173 server) that writes 48 Millions
> lines of data into various text files . The selected data is from various
> tables and I have the query pretty much optimized. Now I am trying to find
> the fastest way to dump the selected data into a text file on the same
> oracle server. The program (written as a pl/sql package) now works
> something like this:
> 
> str     varchar2(32767) := '';
> NL      char(1) := chr(10);      -- new line character
> begin
>  fpn := utl_file.fopen(directory, filename, 'w', 32767);
>  for x in cur1 loop
>    str := str || x.str || NL;       -- keep building the str
>    if (length (str) > 31000 ) then
>       str := substr (str,1, length(str) -1 );
>       utl_file.put_line(fpn, str);
>       str := '';
>    end if;
>  end loop;
> 
>  -- dump the last part:
>  str := substr (str,1, length(str) -1 );
>  utl_file.put_line(fpn, str);
>  utl_file.fflush(fpn);
>  utl_file.fclose(fpn);
> end ;
> 
> 
> The above code works perfect fine now. But I am wondering if there is
> another way that could increase the writing siginificantly faster. This
> porgram does not have to be in pl/sql.  I can think of a couple of
> potential approaches:
> 
> 1. Write a perl program, basically using perl's DBI/DBD to select the data
> from the database, then calling perl's "print" to write data into a file.
> I have not tested this and don't know if it is faster that
> utl_file.put_line.
> 
> 2. Write a C program, using ProC to talk to DB, then use C's fopen and
> fwrite(?) to dump data into text file.
> 
> 3.  Write a C program,using OCI to talk to DB, then use C's fopen and
> fwrite(?) to dump data into text file.
> 
> I don't have direct experience with ProC and OCI, so I don't know how
> faster (or any) it would be  by doing Option 2 or 3 above. Does anyone
> know if I would see siginificant performance boost by using C? Any other
> suggestions?
> 
> TIA.
> 
> Guang
> 
> 
-- Please see the official ORACLE-L FAQ: http://www.orafaq.net -- Author: Tim Gorman INET: tim_at_sagelogix.com Fat City Network Services -- 858-538-5051 http://www.fatcity.com San Diego, California -- Mailing list and web hosting services --------------------------------------------------------------------- To REMOVE yourself from this mailing list, send an E-Mail message to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing).Received on Sat Jan 10 2004 - 22:29:25 CST
|  |  |