I have an Oracle database sitting on a Windows 2000 server. When the the online-redo logs are written to the resources oracle uses goes through the roof and the server grinds to a halt. I presume the LGWR process is not performing efficently but I don't know why. I re-sized the on-line redo logs from 1 to 20 mb but this doesn't seem to have made much difference. Currently the logs are being written about four or five times a day, do you suggest I reduce the size of the logs? Any other ideas would be appreciated.