Chapter 11: Real-World High Availability Solutions

Now that you have learned about high availability technologies, you need to understand how to put the pieces of the puzzle together. This chapter walks you through a real-world example and identifies the decision points and how you would need to confront the challenges. This exercise addresses mainly the technology aspect; the people and process aspect is covered, but is better handled by other examples in this book. The chapter concludes with a case study of Microsoft.com and details of how they achieve high availability.

The Scenario

Your company is starting to plan for a new, custom-built application to support its sales field in North America and Europe. Other global sales territories will be incorporated in later versions. The application will be a sales management tool with other capabilities, such as reporting. The people in the field will be able to access and modify the data from a traditional client/server “based application when they are on the corporate network, and they will be able to view reports from a Web page.

The database portion of the solution must satisfy two main goals that, in the actual solution, can be hosted on more than two physical servers. These goals are to establish:

• A central repository of data

• A reporting server that is updated with five-minute-delayed feeds from the central repository

The company is striving for an availability goal of 99.99%, or fewer than 52 minutes of downtime, for the systems involved with this application. That decision is based on the fact that the systems must service people worldwide, and it factors in all planned downtime (application upgrades, service packs , server upgrades, hardware upgrades, and so on) in addition to unplanned downtime. The solution must be available and scalable and will be in a production capacity for a minimum of two years .

Note

For the sake of this exercise, assume all other parts of the solution, such as domain controllers and Web servers, are accounted for in the overall availability plan but are not factored into cost. In your environments, you have to take into account every last detail of your solution. Also, for this exercise, assume that you have a preferred hardware vendor.