Re: Raw devices vs. File systems for On-Line 5.0

->From: maddah@eggo.usf.edu (Hassan Maddahian) ->Subject: Raw devices vs. File systems for On-Line 5.0 ->Date: 9 Aug 1993 16:57:37 GMT ->Reply-To: maddah@eggo.usf.edu (Hassan Maddahian) ->Organization: University of South Florida, Department of Computer Science and Engineering -> ->Hi, -> ->Question: -> ->I have an ascii data file with 53 fields in it. Each row is composed of 450 ->characters and there are about 400,000 records in the file. -> ->To load the file using dbload into a table with 53 columns, it takes the ->database approximately 100 rows per second with nothing else running on the ->machine. However, If I was to split the table into three smaller tables and ->then load the file into the three separate tables, it takes over 40 hours. -> ->So, why the extra overhead. Would it be faster if I do the load into a file ->system as opposed to a raw device. In addition, when I tried to populate only ->one of the smaller tables at a time, it took an average of only 1 record per ->second being loaded. At this rate, it would had taken over two weeks to get ->it done.! -> ->Any guesses??! -> ->Thankx ->Ali In the three table version, do you by chance have additional indexes, perhaps intended to speed the joining of the tables??? Informix suggests dropping indexes before doing large data loads, and then recreating the indexes afterwards. Makes sense to me: Doing the B-tree rebalancing for 400,000 row inserts could suck your CPU dry! Regards, Alan ___________________________ ______________________| R. Alan Popiel |__________________________ \\ Internet: | Martin Marietta, Tech Ops | / \\ alan@den.mmc.com | P.O. Box 179, M/S 5422 | Std disclaimers apply. / )Voice: | Denver, CO 80201-0179 USA | ( / 303-977-9998 |___________________________| (But you knew that!) \\ /________________________) (____________________________\\