Re: how to speed up SE and DS

> What is your NEXT SIZE for the tables and the row size? Also are the > several tables in the same dbspace? You may be thrashing many > fragments of the several tables interleaved. Run oncheck -pe or > oncheck -pT database:table to check the number of extents (-pe will> also show you how the extents of the various tables are interleaved). > This can be slowing you down. I will check the NEXT SIZE as soon as I am back at the production machine. I have used two dbspaces and I have put the tables in these two dbspaces to allow some load-balancing . But this did not explain the very slow running SE on the LINUX development system. A small joke by the way: SE on Linux runs faster than DS on SUN, both machines are brand new. Because I am converting and loading an old Database (written by me) to INFORMIX, I am able to make some statistic-traces during the import into the database. My program shows how many records are put into the database each minute. Each data-set alters several tables. During the first minute the program loads 247 Datasets, after two hours the program loads only 28 datasets per minute. I have calculated an exponential increase in the time needed for the import with a correlation of more than 99,7%. I have checked, that there is no problem with my loader-program itself, my program processes (loading, parsing and setting up the relational format, because the old database was an OODB-type) the whole load-file (nearly 600.000 lines) in nearly five minutes (if I remove the final call to the ESQL/C-load-function). My program automatically calls UPDATE statistics after every 100 data-sets. -- Andre Koppel Software TCP/IP for QNX 2.x GPIB for QNX4.x at http://www.akso.de and http://www.swd.de