This wiki has been retired. For the latest information on running SQL Server on Windows Azure Virtual machine, see this link: http://msdn.microsoft.com/en-us/library/windowsazure/jj823132.aspx We'll use this wiki to communicate best practices you should consider when provisioning and deploying SQL Server in a Windows Azure Virtual Machine.
Windows Azure Virtual Machine was designed to support a broad range of SQL Server workloads. As with any preview, you should start testing small workloads first and measure throughput and performance. Move on to test more demanding workloads as you gain confidence in the performance, reliability and availability of the service. Consider these guidelines when evaluating which workloads to test. Run some performance tests (seePredeployment I/O Best Practices) to help you compare baseline performance of virtual machines running in Windows Azure Virtual Machine to other platforms.
Big thanks to Sandino Di Mattia for writing these up!
A SQL Server VM in Windows Azure can be added as a replica to an on-premise AlwaysOn Availability Group. This enables a disaster recovery solution for an on-premise SQL Server. If the on-premise SQL Server fails, you can failover the Availability Group to the SQL Server VM replica. In addition, you can configure the SQL Server VM as readable, to offload read workloads and backups. The steps involved are:
Please note the following caveats with regards to application connectivity:
Return to SQL Server in Windows Azure Virtual Machine Early Adoption Cook Book (en-US)
Carsten Siemens edited Revision 45. Comment: fixed typo
Fernando Lugão Veltem edited Revision 36. Comment: removed (en-US) from the title
Richard Mueller edited Revision 34. Comment: Added tag
G Bowerman edited Revision 24. Comment: Fixing reference to OS disk write cache on SQL Server platform images
Richard Mueller edited Revision 17. Comment: Fixed <a name> tag in HTML
Richard Mueller edited Revision 16. Comment: Spelling
Bill Ramos edited Revision 7. Comment: Added warning about default install using the C: drive for the database data, log, and backup files
I have tried to use the failover partner from my app running in Azure Web Sites - I had to open two ports/endpoints 14001 and 14002 in azure so my app could connect to my SQL Cluster running on my VM roles. The problem is once my client connects to the primary it works but I suspect it is receiving a different failover partner as SQL knows nothing about those azure endpoint ports and is sending a different failover partner. It also means my primary and failover server names only differ by port and reading: blogs.msdn.com/.../running-sql-server-default-instance-on-a-non-default-or-non-standard-tcp-port-tips-for-making-application-connectivity-work.aspx -- at the bottom is mentioned this is not going to work.