Hi NG,

I need some advice...

I'm using a DB2 UDB 8.1.2 for Linux.
The database is going to contain about 400 MB of data, distributed over 6 tables.
The database will be reloaded monthly with the LOAD-utility: five tables (about 300 MB) will be completely replaced, a part of the 6. table (about 5 MB from totally 100 MB) will be deleted and then replaced.

I suppose, a simple "create database test" is not enough... Which configuration parameters would be important in this case? What about tablespaces?.. Are 4 MB of log file space enough?..

I do appreciate everyone's help!

Many thanks in advance!

S.B.

When you create the database, use the Create Database Using Wizard on the Control Center. If you can allocate at least 500 MB to DB2 (50% of a 1 GB system) in the Configuration Advisor (see below), I would not worry too much about using database managed space and just used system managed tablespaces for everything. In other words, select low maintenance instead of high performance in the wizard. This is because your entire database will reside in memory (DB2 bufferpool) after it has been read the first time (assuming that your database is 400MB including indexes and overhead).

After the database is created, open the Control Center and right click on the database. Select the Configuration Advisor and answer the questions. This will do a very good job of tuning the parameters.

I would probably use circular logging.

Hi,

thank you very much for your answer!
Unfortunately, the Control Center is not available on the plattform.
Is there another way to tune the parameters?..

S.B.

You can run the Control Center from a remote windows client. It is free from the IBM website (you can install the latest fixpak directly without base DB2 code on windows). The Administrative Client can be downloaded here:
http://www-3.ibm.com/cgi-bin/db2www/...oad.d2w/report

But you can also use the AUTOCONFIGURE command (see the Command Reference manual), which is the same thing in command line mode.

When you create the database, taking the defaults will get you system managed space as I recommended when you bufferpool is large enough to store the entire database.

Hi,

You probalby have KDE or GNOME environment. On Linux command prompt type: kde and press <enter> for kde GUI environment to start.

Hope this helps,
Grofaty

Hi,

Please ensure that you have allocated enough of sort heap and logspace along with the bufferpools as the load command uses these parameters.

If you are loading data that has duplicates then what I have observed is it requires roughly around 2.5 times of logspace of the data to be loaded.

Please also remember to do a set integrity after you have loaded data else you won't get access to tables that share a pk-fk relation.

Hope this helps

Nitin.