Re.: How to tell where to tune.

Andrew, after reading your mail "How to tell where to tune" I would check how much time the online use for checkpoints. If too much time is used too frequently you got a problem. If this is the case I will recommand the following things: 1 Use unbuffered transaction logging instead of buffered. 2 Increase your checkpoint interval to 15 min or maybe 60 min. 3 Start to cleanup your memory when the LRU queues are 10% dirty and stop cleaning at 5% dirty. You may go nearer to the "extreme". My point is, that you may have checkpoints, which take up too much time. The 3 points will give you a smaller peak performance, but should make it less likely, that your user programs "stops" from time to time. Remember to check the physical log for not forcing checkpoint. It may look nicely if all the disk writes are performed at checkpoints, but not all users like, that their programs "stops". Another hint, what about your extents sizes ? The unbuffered transaction logging is not needed for datasecurity on DG's ! ----------------------------------------------------------------- Databaseadministrator Poul Pedersen Kuwait Petroleum (Danmark) A/S Hummeltoftevej 49 Mail: pp@q8.dk 2830 Virum Voice: +45 45 98 45 94 Denmark Fax: +45 42 85 14 18 -----------------------------------------------------------------