Azure storage: staging vs . production

Here is a step by step guide.

1) You need the library: Microsoft.Samples.WindowsAzure.ServiceManagement There is a nuGet package entitled "Windows Azure Service Management Library" which contains this.

2) You need create a X509Certificate2 and follow the instructions laid out here. Make sure you upload the .CER file you create to the Subscription Certificate store. Make sure you upload a copy of the .PFX with the PRIVATE KEY to the actual cloud service certificate store.

Create and upload a certificate for Windows Azure Management

http://blogs.msdn.com/b/clouddeployments/archive/2010/05/12/making-calls-to-the-service-management-api-from-a-service-running-in-windows-azure.aspx

3) These tutorials also gloss over this: you need to have the service endpoint defined. I did that in my app.config file with the following

 <system.serviceModel>
  <bindings>
    <webHttpBinding>
    <binding name="WindowsAzureServiceManagement_WebHttpBinding" closeTimeout="00:01:00"
             openTimeout="00:01:00" receiveTimeout="00:10:00" sendTimeout="00:01:00">
      <readerQuotas maxStringContentLength="1048576" maxBytesPerRead="131072" />
      <security mode="Transport">
        <transport clientCredentialType="Certificate" />
      </security>
    </binding>
  </webHttpBinding>
</bindings>
<client>
  <endpoint name="WindowsAzureEndPoint"
            address="https://management.core.windows.net"
            binding="webHttpBinding" bindingConfiguration="WindowsAzureServiceManagement_WebHttpBinding"                contract="Microsoft.Samples.WindowsAzure.ServiceManagement.IServiceManagement" />     
</client>
</system.serviceModel>

4) Once that is done, I created a static class named "GetServerInstance". Here is the code:

public static class GetServerInstance
  {
      const string SubId = "your azuresubscriptionid";        

    public static bool IsProductionEnvironment()
    {
        //get the current deploymentId
        var currentInstance = RoleEnvironment.DeploymentId;  
        var mgmtChannnel = ServiceManagementHelper.CreateServiceManagementChannel("WindowsAzureEndPoint",GetCertifcate()); //make the endpoint.  
        var serviceDetails = mgmtChannnel.GetHostedServiceWithDetails(SubId, "your-cloud-service-name", true);
        var currentDeploymentSlot = serviceDetails.Deployments.First(p => p.PrivateID == currentInstance).DeploymentSlot;

        if (currentDeploymentSlot == DeploymentSlotType.Staging)            
            return false; //staging server

        if (currentDeploymentSlot == DeploymentSlotType.Production)            
            return true; //production server        

    }


    private static X509Certificate2 GetCertifcate()
    {
        string certificateThumbprint = RoleEnvironment.GetConfigurationSettingValue("CertificateThumbprint");
        if (String.IsNullOrEmpty(certificateThumbprint))
        {
            return null; //I'd throw an exception here and log the error
        }

        var certificateStore = new X509Store(StoreName.My, StoreLocation.LocalMachine);
        certificateStore.Open(OpenFlags.ReadOnly);

        var certs = certificateStore.Certificates.Find(X509FindType.FindByThumbprint, certificateThumbprint, false);
        if (certs.Count != 1)
        {                
            return null; //I'd throw an exception here and log the error
        }

        return certs[0];
    }
}

5) Now in my worker role, which I never want to run on Staging, because it will charge people twice. I call this:

 if (GetServerInstance.IsProductionEnvironment())
        {
           //Do work! I'm in production
        };

This is how I manage my production/acceptance/test environments (note that I'm not using the word staging). For each environment, I create the following (depending on the project):

  • Cloud Service
  • Storage Account
  • SQL Azure Server + Database
  • AppFabric (ACS, ...) Namespace
  • Virtual Machines

So let's assume I have an app called myapp, then my environments would look like this:

  • Production
    • Cloud Service: myapp-prod.cloudapp.net
    • Storage Account: myapp-prod
    • SQL Azure Server containing 1 database: MyApp
  • Acceptance
    • Cloud Service: myapp-acce.cloudapp.net
    • Storage Account: myapp-acce
    • SQL Azure Server containing 1 database: MyAppAcce
  • Test
    • ...

So all environments have a version of the app running in the production deployment slot. I only use the staging deployment slot whenever I want to do a VIP swap for my production environment (note the difference between production deployment slot and production environment).

There are a few advantages to this approach where you have dedicated components (like storage accounts) per environment:

  • It's easy to test new releases without impacting the real application.
  • You can have different security per environment (for example, all developers have access to the keys for the test storage account)
  • If you're testing your application you can work with real URLs + SSL instead of that long and ugly staging URL.
  • It's easy to test the integration with ACS since each environment will have its dedicated namespace.
  • Using Visual Studio you can easiliy manage settings per environment.
  • And last but not least, you have to know that the scalability targets of Windows Azure Storage apply to the storage account level. This means that if you use a single storage account for all your environments, you might be reducing the performance of your app in production because you're doing stress tests on the app running in staging. If you use a storage account per environment, you won't impact other environments when you do something.

Tags:

Storage

Azure

Recent Posts