Windows Azure Guidance – Scenario background – Part I
Here’s the “fiction non-fiction” (I love that term :-)) that will become the backdrop for our initial scenario.
Adatum is a 5000 employee manufacturing company with a large portfolio of applications. Their criticality ranges from “peripheral” to “mission critical”, with lots of “in between. A significant portion of IT budget in Adatum is allocated to “middle” and “peripheral” applications. From an IT perspective, Adatum is mainly a Microsoft technologies shop. Most applications were and are built with Microsoft technologies and tools. They do have however, some legacy systems built on other platforms (e.g. AS400, UNIX boxes, etc).
Adatum’s suppliers and in-house developers are skilled, trained and proficient on the MSFT products, frameworks and tools (SQL Server, Windows Server, .NET, Visual Studio, etc). IT operations and administrators are very familiar with Windows based systems and the tools used to manage them (e.g. System Center, AD, etc).
Adatum wants to optimize its IT portfolio. By “optimize” they mean: “directing their investments on the capabilities that differentiate themselves from their competition” . Adatum is not better than its competitors because they can run e-mail better. However they do differentiate from the competition on other things: better supply chain, better quality controls, etc. Adatum would like to have efficient and effective e-mail but then spend most of their budget on the systems that support these important processes in the company. One way they are considering doing this optimization is by selectively deploying applications to the cloud.
The platform “spectrum”. Giving up control for more economies of scale.
An optimization exercise. Moving capabilities to an optimal environment
Some challenges in Adatum today
Deploying new applications in Adatum today takes too long. Even for simple applications the acquisition, provisioning and deployment processes can take several weeks as the requirements are analyzed, procurement is involved, RFPs are sent to suppliers, networks are configured, etc.
Also, many of their existing applications don’t make the best use of the infrastructure. Many run on underutilized servers, yet they find it difficult to deploy new services on the same hardware with the needed SLAs, making the best use of their infrastructure. In some cases they’ve used virtualization, but not for everything. Also, less critical applications typically get less attention from IT staff, leading to increased availability problems, user satisfaction and higher costs.
Adatum sees the cloud as one way of streamlining these processes. They believe they can leverage the economies of scale of Windows Azure, pushing further standardization and automation throughout.
Goals and some concerns
One concern Adatum has is about user experience of apps moved to the cloud. At minimum, applications should work the same. But it is expected to be better, leading to benefits not just to IT but for the end users. For example, some of their applications would benefit of dynamic scalability, but it is too costly (and slow) for them to acquire resources for peak use. When demand increases, user experience is negatively impacted today. Another example of this, is applications that are currently only accessible on the intranet. Publishing them on the internet has been difficult and users don’t like to VPN. if they are moved to the cloud, they would be automatically available on the internet. However, security then becomes a concern.
And talking about security. many applications today use Windows authentication and users are not required to enter application specific credentials. They wonder how things would work when an application is hosted outside their domain. Also, few applications in Adatum are truly completely isolated. Most have various dependencies to other systems. A lot (meaning: A LOT) of effort goes to integration (a.k.a. HST).
Adatum likes innovation and it is definitely not a laggard, but also wants to take conscious steps when adopting something new. Adatum is open to start evaluating their hypothesis with a few simple applications initially, gain some experience and then expand this approach. One could describe their overall strategy as to “try, learn, fail fast then optimize”.
In general, Adatum, would like to minimize massive retraining of their IT staff. They also want to minimize re-design and re-architecture of their applications, at least initially.
Adatum would also prefer they could keep applications “portable”. Meaning that any given application could be moved back and forth from the cloud to their on-premises data center seamlessly. However, in reality, there’s no truly a need for this on all applications. It is more of a risk perception. There are, however, apps that will stay “portable”.