Special thanks to Sri Harsha for reviewing this post!!

 In this post we will look at:

  • Why Windows Azure updates the host OS?
  • How does the host OS update take place?
  • What are availability sets?
  • How does creating availability sets make your application highly available?
  • Resource to help you create and manage highly available applications in Windows Azure Virtual Machines
  • More Information

 

Why Windows Azure updates the host OS?

Windows Azure deploys updates to the host OS approximately once per month. This ensures that Windows Azure provides a reliable, efficient and secure platform for hosting your applications.

 

How does the host OS update take place?

The host OS update on the Windows Azure platform is different when compared to how you update your PCs or servers running on Windows OS. In case of Windows Azure a new image which contains all the latest updates and fixes is deployed to all the servers and the Fabric Controller will instruct these servers to restart and boot from the newly deployed image. So unlike a Windows update which could take considerable amount of time to complete, the Windows Azure host OS update just involves booting from a new image. Typically this Host OS update process will take 15 to 20 minutes to complete.

 

What are availability sets?

When you have two or more VMs that perform the same task (Example: two or more web servers) you create an availability set with those two VM. Creating this availability makes your application highly available and also makes you eligible for the 99.9% uptime SLA.

 

How does creating availability sets make your application highly available?

When you create an availability set you are instructing the Fabric Controller that all the VM in an availability set perform the same function and  must not be taken down at the same time for scheduled maintenance.

Behind the scene what happens is, the Fabric Controller will intelligently place these VMs on different update domains (UD). These UDs are logical classification which will help the FC to ensure that all VMs in the same AS are not taken down at the same time during any scheduled maintenance. This will ensure that there are VMs which are always available to process requests.

Note:

  1.   Test/monitor to make sure a reduced VM count running the workload provides sufficient performance so your service is not negatively impacted during planned maintenance while one or more VMs are unavailable.
  2. If you are using end point to allow incoming traffic from the outside world ensure that it is load balanced. (See “Creating Highly Available Workloads with Windows Azure” below.)

 

Resource to help you create and manage highly available applications in Windows Azure Virtual Machines

 

More information

Windows Azure Host OS Updates: Why, When and How: http://blogs.technet.com/b/markrussinovich/archive/2012/08/22/3515679.aspx

Role Instance Restarts Due to OS Upgrades: http://blogs.msdn.com/b/kwill/archive/2012/09/19/role-instance-restarts-due-to-os-upgrades.aspx