Editor’s Note: Today’s post comes from Sidney Higa, Principal Programming Writer in our Customer Experience team. This post provides an overview of a very useful paper about assessing costs associated with using Windows Azure for your applications.

 

Are you new to Windows Azure, and someone has asked you to plan and develop an application? "Sure," you reply, "no problem." And, by the way, can you figure out how much it will cost to run, on a monthly basis? That last bit might make take some thinking, especially for someone not familiar with WindowsAzure. But here's one way to approach the problem. We have created a paper, Windows Azure Cost Assessment, which gives you a good starting point. It features a framework and guidance gathered from customers who have already encountered the same situation.

A Solid Starting Point

Based on experience with current customers, we’ve found that there are two major cost centers: Windows Azure Compute, and Windows Azure SQL Database. When using either of those services, charges accumulate 24 hours a day, seven days a week. So given that premise, the paper presents five easy-to-understand-and-implement architectures:

  • Standard Scalable Web Application with Storage
  • Hybrid On-Premises Application Using Windows Azure Storage and SQL Database
  • Background Processing Application
  • Scalable Web Application With Scalable Background Processing
  • Scalable Hybrid Web Application

The architectures show basic configurations of worker roles used in a basic scenario—like running an online storefront and catalog. These architectures are also extensible to accommodate more complex variations. The point is that you can choose one, and start to calculate how many Compute roles you will need to create your basic application.  If you already know how much data resources you will require, you can also plug in that data into the model.

More Info

Given a basic architecture and your own implementation, the paper lists ideas, caveats, and strategies to be aware of as you plan and develop. For example, the basics of capacity planning are given with parameters unique to Windows Azure, such as auto-scaling of worker roles, or testing of low and high usage scenarios, or building-in minimal diagnostic and tracing counters to be used when auto-scaling. (You can only auto-scale effectively if you have some metrics to gauge with.)

With the basic information here, you will have a better grasp on what can seem like a daunting task. And we hope you won’t reach for an aspirin when asked “how much will it cost?”

Check out the paper here.