Setting Service Availability Goals

  • Article

Applies To: Windows Server 2003, Windows Server 2003 with SP1

To set effective service availability goals, you need to measure the percentage of time that the WWW service and your Web sites are available to process client requests. WWW service availability is the time that the WWW service and your Web sites are up and running, expressed as a percentage of a set period of time. For example, an organization might set a goal to have a WWW service availability of 99.999 percent over a 21-day period.

You can achieve high WWW service availability when running IIS 6.0 in worker process isolation mode. This is because worker process isolation mode isolates the WWW service from the impacts of failed applications. Generally, the WWW service is unavailable only when the system is restarted and when IIS is taken offline for major service upgrades, such as the installation of a service pack. When the WWW service is available, your Web sites are available unless you deliberately take them offline.

Test service availability and compare the test results to your goals. First, run the tests in a test environment, and then run the tests periodically in your production environment.

Measuring Service Availability

You can measure WWW service availability by completing the following steps:

  1. Use Performance Monitor to gather the data for WWW service uptime, which measures uptime for W3svc.exe, and uptime for each Web site on the server. The data for WWW service uptime is displayed in the Total instance of the Service Uptime counter. The data for Web site uptime is displayed in the Sites instances of the same counter. For more information about gathering uptime data for the WWW service and Web sites, see Gather and Display WWW Service Uptime Data.

  2. Use Service Monitor to display the data for the Service Uptime counter. The Service Monitor displays the uptime data for the WWW service and for each Web site on the server. For more information about displaying WWW service and Web site uptime data, see Gather and Display WWW Service Uptime Data.

  3. Calculate the availability of the WWW service by dividing the time (in seconds) for the Total instance by the elapsed time that the server was running (in seconds) during the test; multiply the result by 100 to express the figure as a percent availability.

  4. Calculate the availability of each Web site by dividing the time (in seconds) for the Site instance for each site by the elapsed time (in seconds) that the server was running during the test; multiply the result by 100 to express the figure as a percent availability.

  5. Compare the results to your goals.