Figure 1. CPU Utilization Graph
MONITORING SERVER PERFORMANCE
Overview
Servers performance monitoring covers a number of key components, these include;
Network Interfaces
- Mass Storage Devices
- CPUs
- RAM (volatile memory)
Monitoring server components can be achieved using native operating system tools or third party tools that leverage agents or the SNMP protocol. Hardware components are usually associated with a unique SNMP identifier known as an OID. Reading the value of the appropriate OID will allow Network Monitoring systems to graph any SNMP enabled component.
Network Interfaces
Network Interfaces or Network Cards connect the server to network devices and in turn provide accesibility to the servers' resources. Network interfaces (or the corresponding switchport they connect to) should be monitored for utilization because any congestion at the network interface level will impair access to the servers' resources therefore affecting performance. Given the capacity of Gigabit technology today, it is rare for the network interface to be the bottleneck in Server Performance.
Mass Storage Devices
Mass Storage is usually presented in the form of hard drives or hard disks. When users access files on a server they are retreiving information from the hard drives. Some mass storage technologies have been specifically designed for multi-user environments, these include SCSI, SATA and RAID. The server operating system also stores data on the hard drive, this data includes swap files that are used in conjunction with RAM.
CPU
The most common causes of Central Processing Unit high utilization are:
1. Insufficient RAM causing the Operating System to swap large quantities of data to the swap file.
2. Applications that require complex processing such as large database queries or calculating intensive tasks such as image manipulation.
RAM
Random Access Memory is a high performance server component. However, a lack of RAM can cause the hard disk to be used for the overflow memory. Monitoring RAM should show the used memory as well as the available memory.
Tools for Monitoring Server Performance
Native Windows Tools
Windows servers can be monitored using native tools such as Performance Monitor and Task Manager.
Performance Monitor can be configured to monitor in great detail many aspects of the four major sub systems list above (Network Interface, Mass Storage, CPU and RAM). Performance monitor can be configured to store historical performance data, however, by default it will provide real time monitoring only.
Task manager is used primarily for at-a-glance monitoring of server performance. Task manager can report on individual processes, memory and network performance.
Third Party Monitoring Applications
Non-native Windows Server Performance tools use software server agents or SNMP to gather performance data.
Common Windows Server Performance Software include;
- Nagios (Open Source)
- SCOM (Microsoft Corporation)
- Whats Up Gold (Ipswitch Inc)
- Quest Software
- PRTG (Paessler AG)
Related Information
Figure 2. Server Network Interface Utilization Graph
