1. Monitor Server Performance - Monitor the following performance counters
2. Run Away User Processes - Check with the task manager for run away processes.
3. Kill Process - Kill the run away process by logging of the user holding the process or by terminating the process.
4. Memory Usage / User process - Check with the task manager the memory usage and VM size per process.
5. Kill Process - Kill the process if it is consuming too many memory and causing excessive paging.
6. Memory Efficiency - Check the performance counter Cache | Copy Read Hits %. Constant values below 90% indicates a problem with kernel memory.
7. CPU Queue - With performance monitor measure System | Processor Queue Length. This counter represent the number of processes waiting to be processed. A rule of thumb is a maximum of 2 processes per processor.
8. Reduce Processes - Reduce the number of processes by stopping some services or by logging of users from the system.
9. Network Queue - Check the Output Queue Length of the network interface with the performance monitor. If the output queue length is any value above 0 it indicates a problem with the network or network interface.
10. Troubleshoot Network - Troubleshoot the network connections, network load, network interfaces or other network components.
11. Disk Queue - Check the performance counter LogicalDisk | Current Disk Queue Length.
12. Troubleshoot Disk Subsystem - Check the server for failing hard drives. Put pagefile on separate hard drive with it’s own scsi or raid controller.
13. 3rd Level Support - Escalate the problem to a 3rd level support group.
14. Restart Spooler - Restart the spooler service to free up some memory.
15. Stop Kernel Processes -Stop some kernel processes by stopping some services like NetIQ, SNMP, HP/Compaq monitoring agents. This will free up some kernel memory and creates a shorter process list.
16. Reduce Users - Logoff some of the users or add more servers. This will increase the performance.
