Co-sponsors:         

Thank you for coming to the Microsoft Big Day

 

Windows 2000 Server Home

Windows 2000 Server Overview: Reliable

During the design phase of Windows 2000, Microsoft asked customers what they wanted in the next version of the Windows NT operating system. One of the top responses was "increased reliability." In fact, this need drove some potential Windows NT customers to more expensive non-PC platforms for their critical applications. Clearly, making Windows 2000 extremely reliable would be good for customers and good for Microsoft. In response, the developers of Windows 2000 made reliability and availability their top priority.

Finding Opportunities to Improve

The developers began by pinpointing what caused downtime with Windows NT. By analyzing nearly 1,200 servers running Windows NT Server 4.0, the developers learned that 65 percent of reboots were due to planned outages to allow administrators to install and configure hardware, operating systems, and applications. This included cases in which administrators performed "preventative reboots" to try to avoid crashes.

Of the unplanned outages, 21 percent were caused by application failures, and 14 percent were due to system failures. More than half of the system failures were traced to device drivers, anti-virus software, and hardware failures.

In Windows 2000, the development team succeeded in making the OS more reliable and available. This means Windows 2000 is significantly less likely to fail, can continue running during most routine maintenance tasks, and restarts faster after a reboot.

Built to Keep Running

The first step toward improving reliability was to change the way the developers design, develop, and test Windows. Microsoft introduced new development tools and dedicated reliability teams, and conducted nightly system-wide stress tests that simulated millions of users. This refined process let developers better track system failures and monitor planned downtime for maintenance that reduced system availability.

Among the culprits for systems failures: faulty driver software. So the developers conducted tests with anti-virus software and driver software developers. To prevent crashes in the future, independent software developers and hardware vendors can now test their code using the Windows 2000 Driver Verifier tool.

To help keep systems up and running, Windows 2000 greatly reduces the number of maintenance tasks that require rebooting the computer. Now, you can install most hardware and software or adjust many configuration settings without rebooting.

Other reliability tools: A resource-partitioning feature prevents application failures from forcing reboots, and an improved Task Manager lets administrators kill entire process trees to completely shut down a "misbehaving" application.

Lastly, to reduce the amount of time systems are offline, when a system fails or is taken down for maintenance, new boot options let administrators quickly restart the system.

Advanced Reliability

Many organizations will use Windows 2000 for e-Commerce and customer relationship management to take advantage of its extensive support for custom Web and application development. These applications can take advantage of the added reliability offered by Windows 2000 Advanced Server. This version provides clustering and load-balancing technologies that allow two server PCs to handle the load of a single application. If one of the servers fails, the other can assume the load so the application keeps running.

For the largest applications, Microsoft offers Windows Datacenter Server, which is designed for enterprises that need high-end, very reliable hardware and software for high-traffic networks.

Conclusion

The Windows 2000 Server product line is the most reliable set of server operating systems Microsoft has ever produced. The reliability improvements in Windows 2000 mean fewer network interruptions for end users, higher server uptime, and better availability. Further, Windows 2000 Server, Advanced Server, and Datacenter Server provide increasing levels of reliability, to let the operating system readily support high-traffic Web sites, high-volume transaction processing, and many other demanding applications.