Microsoft Azure is designed for cattle type applications. We discussed cattle versus pets in Chapter 1, An Introduction to Cloud Computing.

Azure does not have facilities to move virtual machines without downtime to other nodes in case of planned maintenance. Virtual machines will restart when Microsoft installs updates on the nodes. For this reason, Microsoft does not have a Service Level Agreement for single-instance virtual machines.

For customers to get guarantees on availability via an SLA, there needs to be at least two instances serving the same application role that are required to be part of the same availability set. In the next chapter, we will discuss availability sets.

Microsoft also does not support the Windows Server Failover Clustering role in virtual machines running on Azure. The only exception is SQL Server AlwaysOn Availability Groups, which are fully supported by Microsoft. So Windows clustering cannot be used to have redundant instances...