Performance monitoring and Scalability Tools (IIS, operating system)

The monitoring server is a critical part of server management. By using the appropriate monitoring tools, you can detect server problems, evaluate the results of changes to site content, and plan upgrades to facilitate user access to the site. Select the appropriate monitoring tools and methods based on the information you need.

You can also monitor the server by checking the logs generated by IIS. These logs extend the logs generated by Microsoft Windows. For more information about logging, see Logging Site activity.

This topic includes information about the following monitoring tools: System Monitor Performance Logs and Alerts Event Viewer Task Manager Network Monitor Log Manager WEB Application Pressure Tools Application Center Test tool System Monitor

Windows Performance monitoring Tool, System Monitor (formerly known as Performance Monitor), is a Microsoft management console (MMC) snap-in that you can use to monitor the activity of the server and summarize its performance at selected intervals. With this tool, you can display performance data in a live chart or report, collect data from a file, and generate a warning when critical events occur. Data can be used to determine the cause of system bottlenecks and to adjust system and application performance. Performance Bottlenecks

Bottlenecks are a point of slow data transfer in a server environment. Bottlenecks are often caused by insufficient resources, such as memory or CPU, or due to improper configuration of software, hardware, and network resources. For example, a processor bottleneck occurs when a process starts a backup because the server does not have enough processor cycles to effectively complete a number of requests. Poor application design is also a common cause of bottlenecks. Poorly designed applications can affect servers such as IIS, which can handle thousands of requests per second, and reduce processing power to a small portion of the server's capacity.

To resolve bottlenecks, you must monitor your server environment with the aid of performance counters and monitoring tools such as System Monitor, and collect data about specific components and resources, such as CPUs, memory, applications, and hardware. Analyzing the monitoring data may uncover problems, such as excessive demand for some resources or components that can lead to bottlenecks. Performance Counters

System Monitor can check the output of performance counters. Performance counters can monitor the activity of specific performance objects. In other words, performance counters monitor specific services or mechanisms that control server resources. For example, if you click Add (Plus) on the System Monitor toolbar, you can view a list of performance objects from the performance objects list. Click the Processor object to view a list of the corresponding counters, such as the % Processor Time counter. You can add or remove counters from this window in the graphical display of System Monitor, or configure counters to be logged in a specific way. Running many counters at the same time can result in a slight performance degradation and therefore a choice.

members of the Microsoft Windows Server 2003 family include hundreds of counters. IIS installs its own counters, including the World Wide Web Publishing Service (WWW service), the WWW service cache, the FTP service, and Microsoft Active Server Page (ASP). WWW Service counters, FTP service counters, and ASP object counters can monitor the connection activity of scripts. For a detailed list of performance-related counters, see performance counter references.

System Monitor is used to provide short-term and long-term information about the load on the server and the site. The following table lists some useful counters and provides an ideal average for each counter. System Monitor charts are plotted with a scale of 0 to 100 (by default), so you may need to change the scale of individual counters that record smaller values.

Objects/Counters	Ideal Value
memory/pages/sec	0-20 (if greater than 80, indicates a problem).
memory/Available Bytes	At least MB.
memory/committed Bytes	No more than 75% of the physical memory size
memory/Pool nonpaged Bytes	Stable (slow growth indicates a memory leak)
processor/% Processor time	Less than 75%
processor/System Processor Queue Length	4 or less than 4
(Logical or physical) disk/% Disk time	As low as possible
(Logical or physical) Disk/avg.disk Queue Length	Less than 4
(Logical or physical) disk/AVG Disk bytes/transfer	As high as possible
system/ContextSwitches/sec	Monitor and ensure that this number is lower (relative to System calls/sec values).
system/System calls/sec	As low as possible
WWW service/Bytes total/sec	As high as possible
WWW Service cache/File cache Hits%	As high as possible (static content) Note This value may be low if the Kernel URI Cache Hits is high.
WWW Service cache/Kernel:uri Cache Flushes	Monitor and ensure that this number is lower (relative to the number of requests). Note This number will be larger whenever a file is refreshed from the HTTP.sys answer cache (HTTP.sys is a kernel-mode device driver in IIS 6.0), which means that the content has not been accessed in the last 2-4 minutes. The only way to reduce this counter is not to flush the cache too frequently, although doing so may mean that HTTP.sys uses more memory on content that is not accessed or obsolete.
WWW Service cache/Kernel:uri Cache misses	As low as possible
WWW Service cache/Kernel:uri cache Hits%	As high as possible (static content)
Active Server pages/ Request wait Time	As low as possible
Active Server pages/ Requests Queued	As low as possible
Active Server pages/transactions/sec	As high as possible Attention ASP transactions degrade server overall performance because each transaction needs to interact with the database. If you pay attention to the performance of your computer, you should use less ASP transactions.

For more information, search for "System Monitor" in Windows help. Performance Logs and Alerts

Performance Logs and Alerts contain features for recording counter and event trace data and for generating performance alerts. Counter logs allow you to record data about hardware usage and system service activity from a local or remote computer. Records can be performed manually or on a user-defined schedule, as needed. You can also use continuous records, which are subject to file size or duration constraints. The recorded data can be viewed using System Monitor display, or it can be exported to a spreadsheet program or database for analysis and report generation.

trace logs record data as certain activities, such as disk I/O or page faults, occur. When an event occurs, the provider sends data to the log service. For more information about this tool, see Capacity Planning tracking. Event Viewer

The Event Log service logs events such as an error or a successful service startup. You can use Event Viewer to view event logs. You can also use Event Viewer to monitor system, security, and application event logs. You can use this information to better understand the order and type of events that cause specific performance problems. For more information about this tool, search for Event Viewer in Windows Help. Task Manager

Task Manager displays real-time network usage data (including CPU and memory usage) for various network interfaces on your computer. You can use Task Manager to view the tasks and threads that are in progress, to change the priority of process assignments, and to monitor network and user information. For more information about this tool, search for "task Manager" in Windows help. Network Monitor

Network Monitor captures network traffic information and provides detailed information about the frames that are being sent and received. This tool can help you analyze complex network traffic patterns. Network Monitor can help you view header information that is included in HTTP and FTP requests. Typically you need to design a capture filter, which is similar to a database query and picks out the frames that are being transferred. You can also use capture triggers, which respond to events on the network by initiating an action, such as starting an executable file. A simplified version of Network Monitor is included in members of the Windows Server 2003 family. Microsoft Systems Management Server includes a full version of Network Monitor. For more information about this tool, search for "Network Monitor" in Windows help. Log Manager

when designing a Web site, it is important to consider future growth and understand which parts of the IIS subsystem within the Windows Server 2003 family may be affected by the increased load. This is called "Capacity planning." For information about how to use the Log Manager to plan capacity, see Capacity Planning tracking. WEB Application pressure tools

The Web Application Stress tool is used to realistically simulate multiple browsers requesting pages from a WEB application at the same time. The Web Application Stress tool can be used to generate customizable workloads on a variety of Internet services. The WEB application pressure tool has a limited number of URLs that can be pressured at the same time (about 100). Therefore, when testing a large number of sites, the WEB Application Stress tool may need to be used with other tools, such as the Microsoft Web capacity Analysis tool. The Web Application Stress tool is available in the latest version of the Windows Resource Kit included with the CD, or it can be found in the Microsoft Web application Stress Tool Web site. The Microsoft Web site also provides information about the test tools, including scenarios and results. For more information, see Performance Testing with the Web application Stress Tool. Application Center Test

Microsoft Application Center Enterprise Edition includes a variety of high-end server monitoring tools, such as Application Center Test.

---additional Chinese counter instructions---

Objects/Counters	Recommended values
Memory/Pages/sec	00-20 (if greater than 80, indicates a problem).
Number of memory/free bytes	At least 4 MB.
bytes of Memory/committed	No more than 75% of the physical memory.
Memory/nonpaged Byte pool	Fixed (slow growth indicates a memory leak problem).
Processor/% Processor Time	Less than 75%.
Processor/Interrupt Count/sec	is related to the processor. The maximum 486/66 processor is 1,000;P90 for 3,500;P200 is greater than 7,000. The lower the better.
Processor/Processor Queue Length	Less than 2.
Disk (Logical disk or physical disk)/% Disk Time	as low as possible.
Disk (Logical disk or physical disk)/Queue Length	Less than 2.
Disk (logical or Physical disk)/Average disk byte/transfer	As high as possible.
Global Internet Information Services/cache Hits%	As high as possible.
Web Service/Total bytes per second	As high as possible.
Active Server pages/Request Wait Time	as low as possible.
Active Server pages/Queued Request	Zero.
Active Server pages/Transactions/sec	As high as possible.