Scenario : Have single server running SQL 2000 Enterprise on a RAID 5 configuration. There is only one application which connects to this instance and there are 7 connections spread across the 5 databases defined. The connections are kept open to the DB. Each database is set to SIMPLE recovery. While using perm monitor. counters, I notice the following. 1) Checkpoint pages/sec goes from 0 to 600/sec for about 6 seconds. 2) Physical page writes/sec follows the same as the the checkpoint. 3) Physical disk writes shows about 3 MB/sec during this time. 4) Percent log used on one of the DB reached 70% then drops to 20%. 5) Load is low in my opinion, averages 10 batch requests / second. 6) This is much more pronouced on the RAID 5 boxes in comparison to the IDE test boxes, it appears this is related to write caching? 7) When this occurrs our time critical app is slowed drastically. This matches the documentation on log being truncated at 70% during a checkpoint on the database right ? So, I thought I could force a checkpoint more often in an attempt to reduce the amount of writes. I did this by scheduling a job to run every 2 minutes( test value) that simply called 'CHECKPOINT'. Notes: This database file is 100 MB and the LogFile is 50 MB. Also, this database only has 'INSERTS' going against the DB 95% of the time. I am looking into the possibility of using a write cache on the controller if even available. I don't need 50 MB for the log file, if I shrink it will that help?