This post is taking on the next set of questions we have seen over the past weeks.

What is the model for Interoperability

The Windows Azure Platform is built for cross-platform interoperability, with a commitment to support SOAP and REST, XML, Java, PHP, Python and Ruby.

Can I run applications built for Windows Azure Platform on servers running Windows Server in my own datacenter?

  • At the current time some Windows Azure ASP.NET and PHP applications leveraging SQL Azure can be easily ported to run on Windows Server.  However, Windows Azure applications that take advantage of the unique Windows Azure APIs and blob storage feature will not be able to run on Windows Server without significant re-factoring. In the future Microsoft will provide greater symmetry of application platforms across Windows Server and Windows Azure. As one example, Microsoft’s new application server code-named “Dublin” will be released for Windows Server and then will run on Windows Azure in the future.

Partner Model

Below are the 2 models we currently envision as possibilities for Hosters to build on or with Azure. When a Microsoft partner procures the Azure services through the Microsoft Partner Network Portal a 5 percent discount is applied on Windows Azure platform services, except data transfers and Windows Azure storage.

The Embedded model:

clip_image001

This model can be used to create new, innovative services where Azure is integrated in the back end, like your current server based infrastructure. Hosters don’t (have to) disclose the use of Windows Azure and maintain all end customer relationship on billing and support.

The ‘Build’ for Azure model:

clip_image002

The ‘Build for Azure’ model allows Hosters to sell custom applications as part of a larger bundled solution, where the end-customer signs up for the procurement of Windows Azure services direct.

I want to consider running some of my customer workloads on the Windows Azure Platform. What do I need to do to build and resell a Windows Azure platform-based hosting service to my customers?

  • The first choice you must make is whether you wish to bill the customer for their Windows Azure platform usage yourself or have Microsoft bill them for you. Microsoft has two partner models, an ‘Embedded’ Windows Azure platform model and a ‘Built For’ Windows Azure platform model.   If you choose the ‘Embedded’ model, then you will have your own account with Microsoft, you will build tools and utilities on top of the Windows Azure platform that your customers will access to make use of the platform, and any resulting usage will be billed to your account.  

If you want to map your aggregate Windows Azure platform usage back to individual end customers for rebilling purposes, it is up to you to include the appropriate usage tracking in your tools and utilities. If you choose the ‘Built For’ model, your end customer will sign up for a Windows Azure platform account directly with Microsoft, and then access your tools and utilities after they have logged in so that usage is billed directly to their own Windows Azure platform account. In this scenario you have no involvement in the Windows Azure platform billing but may charge additional fees for the use of your tools and utilities. In either case, your job is to provide the custom screens, tools and utilities that your customers will use to simplify and streamline their use of the Windows Azure platform and make it appropriate to their needs.   

How do I provision Windows Azure Platform accounts on behalf of my end customers, give them access and receive their usage bills without the customer directly interacting with Microsoft?   

  • Today the Windows Azure Platform does not have a publicly accessible, programmable account management and billing API. All account provisioning, logon and usage reporting are managed through a web user interface. Partners, therefore, have an opportunity to build out their own provisioning and account management layer that supports all their customers within a single Windows Azure platform account owned and controlled by the partner.  

The security necessary to segregate one end customer’s usage of the platform from another within the same billing account is readily available in the Windows Azure and .NET Services APIs and in the SQL Azure database offering. This is in fact the standard multi-tenant approach used by major Software-as-a-Service ISVs in their applications today. What a hoster must do if they wish to re-bill underlying Windows Azure platform usage (expressed in service hours, storage and network bandwidth) directly to their customers is develop the tools necessary to track and record their individual customers’ usage of the Windows Azure Platform services.    

Time to market

The EMEA Datacenter locations will go live by January 4th 2010

Usage of Windows Azure services in January will be free of charge. Users will receive a ‘zero bill’ end of January to understand the costs for their application usage pattern.

How do BPOS and Azure relate

  • The current Business Productivity Online Suite (Exchange Online, SharePoint Online, Office Communications Online, and Live Meeting) was built and released before the introduction of the Windows Azure platform. While the two share common operational elements such as datacenters, provisioning and identity infrastructure, and underlying commerce platform,  the initial BPOS services were created based on the current releases of each server e.g. Exchange 2007.  You should expect that over time, we will further align our investments in Microsoft Online Services to take advantage of the scale and flexibility offered by Windows Azure, SQL Azure and .NET Services.

What is the SLA and how is this monitored?

Windows Azure has separate SLA’s for compute and storage. For compute, we guarantee that when you deploy two or more role instances in different fault and upgrade domains your Internet facing roles will have external connectivity at least 99.95% of the time. Additionally, we will monitor all of your individual role instances and guarantee that 99.9% of the time we will detect within two minutes when a role instance’s process is not running and initiate corrective action.