by Jay Milne  Server Performance: Benchmarking, Monitoring and Avoiding Bottlenecks Monitoring Tools A number of tools are available that can help you identify which bottlenecks are the biggest problems. The process of identifying the bottleneck(s) is more of an art than a science. Each commercial operating system provides some set of tools: NT has its Performance Monitor, and NetWare has the Monitor NLM. For the most part, these OS-provided tools provide a basic level of functionality. To get more detailed information and features, you'll have to turn to third-party tools. A number of third-party tools help you to better understand the performance of your server. The most important characteristic of any tool you use is the impact that it has on the system under test. Ideally there should be no impact, but that is not a reasonable assumption. The faster the server the less im pact the tool will have, but you need to take into account the system overhead imposed. This can come in the form of CPU cycles, network bandwidth usage and memory usage. If you are running Microsoft NT and using Microsoft BackOffice applications, a product from Net-IQ provides a good set of tools to monitor the performance of NT Server and BackOffice applications. For those in a mixed NetWare-NT environment, HawkNet provides a product that monitors both platforms-thus avoiding running multiple consoles. In the Unix environment, many administrators use the tools that come with the core operating system. Utilities such as ps and vmstat are commonly used. For Sun Solaris, there is a popular utility called iostat that gives good information about I/O performance. Any performance tool should be able to let you measure the performance of your server over time. Taking just one or a few samples is not enough. You need to measure the performance over a period of time to give you a baseline to measure against. As you place additional resources on the server, such as more users or more applications, you will be able to see how your server reacts. In the PC-LAN world, servers are often looked upon as black boxes. Only a limited set of tools is available to view what is happening at the operating system level and at the hardware level. Without having the knowledge of what exactly the server was doing, you have no real good idea of how to fix the problem. To help solve this dilemma, many of the hardware vendors have instrumented their SCSI adapters, disk drives, network adapters and motherboards. And when used with the appropriate software, you can get a better look inside the server. What kind of information do you get? Compaq's Insight Manager software, when used with any Compaq server, lets you see EISA and PCI bus utilization, SCSI bus utilization, SCSI bus errors, and disk errors, all of which are critical in determining the health of your server. Server Performance Benchmarking Bottlenecks Common Tips and Tricks Updated May 21, 1997 Print This Page E-mail this URL