Important Cloud Services (classic) is now deprecated

Swap deployments to promote a staged deployment to production

When you decide to deploy a new release of a cloud service, stage and test your new release in your cloud service staging environment. Use Swap to switch the URLs by which the two deployments are addressed and promote a new release to production.
You can swap deployments from the Cloud Services page or the dashboard.

1. 
   In the Azure portal, select the cloud service you want to update. This step opens the cloud service instance blade.
   
2. 
   On the blade, select Swap.
   

3. 
   The following confirmation prompt opens:
   

4. 
   After you verify the deployment information, select OK to swap the deployments.
   

The deployment swap happens quickly because the only thing that changes is the virtual IP addresses (VIPs) for the deployments.
To save compute costs, you can delete the staging deployment after you verify that your production deployment is working as expected.

Common questions about swapping deployments

What are the prerequisites for swapping deployments?
There are two key prerequisites for a successful deployment swap:

• 
  If you want to use a static IP address for your production slot, you must reserve one for your staging slot as well. Otherwise, the swap fails.
  
• 
  All instances of your roles must be running before you can perform the swap. You can check the status of your instances on the Overview blade of the Azure portal. Alternatively, you can use the Get-AzureRole command in Windows PowerShell.
  

Note that guest OS updates and service healing operations also can cause deployment swaps to fail. For more information, see Troubleshoot cloud service deployment problems.
Does a swap incur downtime for my application? How should I handle it?
As described in the previous section, a deployment swap is typically fast because it's just a configuration change in the Azure load balancer. In some cases, it can take 10 or more seconds and result in transient connection failures. To limit impact to your customers, consider implementing client retry logic.