Achieving 99.999 Uptime Using Windows Server 2003

Achieving 99.999% Uptime Using Windows Server 2003

When the topic of disaster recovery comes up, many people think of the phrase "five nines" or "99.999% uptime." Although understanding this concept is reasonably simple, actually providing five nines for a server or a network can be quite a large and expensive task. Achieving 99.999% uptime means that the server, application, network, or whatever is supposed to have this amount of uptime can only be down for just over five minutes a year. Having such success is quite a claim to make, so administrators should make it with caution and document it, citing explicitly what this service depends on. For example, if a power failure occurs and the battery backups will last only two hours, a dependency for a server could be that if a power outage occurs, it can withstand up to two hours without power.

To provide 99.999% uptime for services available on Windows Server 2003, administrators can build in redundancy and replication on a data, service, server, or site level. Many Windows Server 2003 services outlined in previous chapters in this book, including Cluster Services, network load balancing, the Distributed File System, and the File Replication Service, can provide redundancy at each level described here.

Providing Redundant Domain Services

To provide redundant domain and network services to be able to deliver five nines uptime, administrators can use Windows Server 2003 built-in services and clever designs to meet the particular type of failure or disaster they are planning for. For example, if an organization provides a Web-based Internet application, five nines can be achieved in a single location by deploying the application on a network load-balancing cluster of servers. To provide this same application across separate sites, the organization would need to establish a VPN across the sites so that a single DNS record could be used to access the server from either location.