Slow system ?! - using Informix 7.x on Sun Sparc 1000

Hi, My company has a customer which is running our app (uses informix for RDBMS) on a Sparc 1000. The system has the following specs: - 4 CPU's (don't know CPU speed), 512 MB RAM, 8 GIG of HD (6 GIG for informix database, 3 2.1 GB HDs). - This system averages 600,000 transactions per day. Here are some quick "onconfig" parameters: - rootdbs is 500 MB - physfile is 10 MB - logsize is 10 MB and there are 5 of them - initial shmem 16 MB, additional shmem 16 MB, no bound set - checkpoint interval is 300 seconds We are experiencing a problem where the system runs slow. After a quick examination, it was seen that under normal operation is checkpoint was being performed about every 3 minutes with each one lasting about 45 seconds. This appears to be the major problem as to why database access is slow. I appears that the physfile space is filling up too fast and reaches the 75% threshold level too quick. So, I am thinking of raising it from 10 MB to 50 MB (1/10 of the rootdbs where it resides). Does this seem reasonable? Are there any general guidelines as to the sizes of both physical and logical logs (without having to figure out userthreads and critical sections, etc..)? Also, are there any other ways to reduce the amount of time it takes to perform a checkpoint? Gauging from other systems that even have higher volume, a checkpoint should only take a couple of seconds! Any ideas would be appreciated. On a related topic, does anyone recommend the informix classes on tuning Online and in your opinion would a person be able to configure and tune a large scale system running informix after taking such a course (assuming that person had practical work experience using informix)?? Thank You George Kydes.