Suppose you have an existing application that you want to migrate to the cloud. There are many things to look at, from the data to the architecture to considering how you plan to scale beyond one instance. For simple web sites which don’t use a lot of data or stateful information the migration to the cloud can be pretty easy with the installation of the Visual Studio Tools for Windows Azure.
The current release of the tools (as of March 2012) is v1.6 which adds a command to the context menu when you right click on a web project. The SDK is installed using the Web Platform Installer (aka WebPI) which checks for previous versions of the tools and any other prerequisites and the consolidates the install into a single step.
After installing the tools when you open a Web Project, whether ASP.NET Web Forms, MVC, or similar, right clicking the project file gives you the command to add an Azure Deployment Project. You could achieve something similar by simply adding a Cloud Project to your solution and then right clicking on the “Roles” folder and adding the existing site to the project.
This adds a Cloud project to the solution and then adds the current Web project as a standard Web Role in the deployment. The key assets added include a Service Definition file and a couple Service Configuration files (one for running locally and one for running in the cloud).
The Service Definition file includes markup which describes how the Cloud Service will be deployed, including the public ports that are enrolled in the load balancer, any startup tasks and plugins that you want to run, as well as any other custom configuration that should be part of the deployment.
For example the Service Definition we created for the Kick Start event in Minneapolis looked like this:
<?xml version="1.0" encoding="utf-8"?> <ServiceDefinition name="MplsKickSite.Azure" xmlns="http://schemas.microsoft.com/ServiceHosting/2008/10/ServiceDefinition"> <WebRole name="MplsKickSite" vmsize="Small"> <Sites> <Site name="Web"> <Bindings> <Binding name="Endpoint1" endpointName="Endpoint1" /> </Bindings> </Site> </Sites> <Endpoints> <InputEndpoint name="Endpoint1" protocol="http" port="80" /> </Endpoints> <Imports> <Import moduleName="Diagnostics" /> </Imports> </WebRole> </ServiceDefinition>
In this code we’ve got a single web role being deployed that will run on port 80. It is running a small instance (1 dedicated core), and will have remote desktop enabled. The Service Configuration contains additional settings including the OS Family for the instance (1 = Windows Server 2008 SP2 and 2 = Server 2008 R2) and number of instances to run (in this case 3).
<?xml version="1.0" encoding="utf-8"?> <ServiceConfiguration serviceName="MplsKickSite.Azure" xmlns="http://schemas.microsoft.com/ServiceHosting/2008/10/ServiceConfiguration"
osFamily="2" osVersion="*"> <Role name="MplsKickSite"> <Instances count="3" /> <ConfigurationSettings> <Setting name="Microsoft.WindowsAzure.Plugins.Diagnostics.ConnectionString" value="UseDevelopmentStorage=true" /> <Setting name="Microsoft.WindowsAzure.Plugins.RemoteAccess.Enabled" value="true" /> <Setting name="Microsoft.WindowsAzure.Plugins.RemoteAccess.AccountUsername" value="... <Setting name="Microsoft.WindowsAzure.Plugins.RemoteAccess.AccountEncryptedPassword" value="...
<Setting name="Microsoft.WindowsAzure.Plugins.RemoteAccess.AccountExpiration" value="... <Setting name="Microsoft.WindowsAzure.Plugins.RemoteForwarder.Enabled" value="true" /> <Setting name="QueueName" value="createthumbnail" /> <Setting name="myStorageAcct" value="UseDevelopmentStorage=true" /> </ConfigurationSettings> <Certificates> <Certificate name="Microsoft.WindowsAzure.Plugins.RemoteAccess.PasswordEncryption" thumbprint="... </Certificates> </Role> </ServiceConfiguration>
Pressing F5 will run the Cloud Service in the local emulator (assuming we’re running Visual Studio as Administrator so it can communication cross process to the running services…an error message tells you if you’re not). The nice thing about this approach is it supports attaching to the local debugger and using the rich set of features such as breakpoints, local and watch variables, and all the rest of the goodness we’ve come to expect in Visual Studio debugging.
Once we’re satisfied that the application works and we are ready to go to the Cloud, you can right click on the Azure Deployment project and either click “Package” to create a deployment package which you can manually upload the the Windows Azure Management portal, or you can select Publish to automate it directly from Visual Studio. This runs a wizard with 3 basic steps to collect the necessary information and complete the deployment. You select the subscription you want to use, name the hosted service, optionally enable remote desktop and web deploy, and then click complete.
Note that if you enable Web Deploy, because this is only supported for a single instance development environment where you require the ability to publish changes to a web site quickly, the publishing process will change your configured number of instances to 1 despite having set it to 3 in the configuration files. That’s because if you have more than 1 instance deployed the load balancer will distribute the incoming web requests round robin, and if you change one of the instances with a web deploy you’ll have an inconsistent state. You can always change the number of instances from the management portal, but you should be aware of what you’re doing.
What happens next is the deployment process takes place. This includes validating that the hosted service is provisioned for you, uploading certificates if needed, then uploading your deployment package and configuration files to Windows Azure storage, and then deploying and starting your instances. It will also add the service deployment to the Server Explorer window (CTRL + ALT + S). From there you can see the status of your service instances, download the profile or intellitrace files if you’ve enabled that option (under advanced features during the publish).
For more information on how to do this, or if you’d like to see a webcast on how this is done, check out the first webcast in the Cloud Computing Soup to Nuts series (http://bit.ly/s2nCloud).