What is Windows Azure and Infrastructure as a Service?
Windows Azure is a set of Microsoft public cloud services. It represents a trend where capabilities that can be provided by a cloud vendor are configured and consumed by customers. These services can be consumed in one of two ways: as a standalone cloud-based implementation or as a hybrid of cloud-based and on-premises services. Infrastructure as a service (IaaS) is not a Microsoft service offering, but it represents a step on the continuum of on-premises deployments (where everything is managed by the customer) to software-as-a-service (SaaS) deployments (where everything is managed by the service provider):

Figure 1. Services Differentiated by Customer/Provider Responsibilities

Source: http://mythoughtsonit.com/2011/04/infrastructure-as-a-service-platform-as-a-service-software-as-a-servicetake-a-look-at-the-management-stack/

IaaS services include the following:

The cloud delivery model allows Microsoft to update and add new services more quickly than what can be done for on-premises customers. One way that Windows Azure reflects this paradigm is with a frequent cadence of updates. (These updates are not limited to IaaS components.) Server images are updated. New services (like the cache service) are previewed and, if things work out, made generally available after the preview period.

The business drivers for IaaS reflect the challenges that customers have in deploying and managing infrastructure components:

  • Speed
  • Cost
  • Governance (a.k.a. "not in my datacenter")

Our customers want to know how they can take advantage of IaaS. The road to this knowledge includes the following stops:

  • Awareness (What is IaaS?)
  • Use cases (What are some things that we do today that we could do in IaaS and what would that look like?)
  • Possibilities (What can we do with IaaS that we can't do today?)
  • Benefits (Why is this better than what we do today?)

Why Microsoft?
Common workloads like Microsoft SharePoint were among the first candidates for IaaS among our customers. On-premises deployments of SharePoint have historically been challenging to upgrade because SharePoint is frequently classified as a business-critical application and the downtime required by an upgrade or migration can affect productivity. Even though capabilities delivered by the latest version of SharePoint are often in high demand by customers, it can be challenging to simply set up a demonstration environment.

Our response, at the Microsoft Technology Center (MTC) in Detroit, Michigan, to the SharePoint dilemma is to demonstrate the process of deploying a SharePoint farm in Windows Azure. I wrote this blog post to demonstrate how IaaS can address the aforementioned business drivers:

  • Speed – The manual process of deploying a network, domain controller, Microsoft SQL Server, and SharePoint Server, in addition to exposing that SharePoint Server to external users, takes fewer than three hours.
  • Cost – Choosing one medium virtual machine (for the domain controller) and two extra-large virtual machines (for SQL Server and SharePoint) costs approximately US$1300 per month to run the servers 24x7.
  • Governance – It’s not in the customer’s data center, but it can be managed like it is.

The simplest SharePoint farm requires three servers:

Figure 2. The minimum number of servers required to deploy SharePoint Server with a separate SQL Server.

This provides a great example of how IaaS provides building blocks that can allow customers to move quickly. For the three servers in the SharePoint farm, there are three server images that can be used with no modification and minimal configuration:

  • Windows Server 2012 R2 Datacenter (Domain Controller)
  • Microsoft SQL Server 2012 Enterprise (SQL Server)
  • Microsoft SharePoint Server 2013 Trial (SharePoint Server)

The following figure shows some of the images that are provided:

Figure 3. Images provided in IaaS (sorted by category)

IaaS doesn’t just provide building blocks; it also provides different types and sizes of the components used to create an infrastructure. Standard instance virtual machines run from extra small (A0) with shared virtual cores and 768 megabytes (MB) of RAM all the way to extra large (A4) with 8 virtual cores and 14 gigabytes (GB) of RAM. For memory intensive applications, the A5 instance provides 2 cores and 14 GB of RAM, A6 provides 4 cores and 28 GB of RAM, and A7 provides 8 cores and 56 GB of RAM.

This combination of predefined server images and a wide variety of virtual machine configurations provides flexibility beyond what many enterprise customers have in their own data centers.

Why Now?
IT organizations’ business customers have come to expect a turnkey experience for any commodity service. SaaS applications like Microsoft Office 365 have raised the bar in terms of capabilities delivered and the amount of time required to deploy and configure these capabilities. (SaaS has also raised the bar for what’s considered a “commodity service”.) Current procurement processes, restrictions around virtual machine sizes and operating systems, and even internal politics are no longer sufficient justifications for telling a business customer that it’s going to take six months for their servers to be configured.

Other challenges faced by IT include:

  • Testing software updates
  • Custom development
  • Deploying servers with non-standard operating systems

Developing custom solutions for SharePoint is an example of this. An IaaS solution is a standalone instance of SharePoint (including all the requisite development tools) that runs outside the corporate network and is accessible via Remote Desktop Connection.

Our customers are interested in IaaS because their current model (on-premises deployments with on-premises security requirements, procedures, and policies) may not allow them to address business issues effectively. IaaS is not a panacea. It is a robust, constantly evolving set of infrastructure components that can be leveraged as another tool in the IT pro’s toolbox.


Joe Baeza is a Collaboration Technical Architect at the Microsoft Technology Center (MTC) in Detroit, Michigan. He has worked for Microsoft for 16 years. The work Joe does at the MTC is challenging because technology keeps changing, frustrating for the same reason, and exciting because he works with so many different customers across so many different industries.