Moving to a software-defined data center: Think big, but start small

You’d like to move toward a software-defined data center (SDDC). So where do you begin? Aaron Bawcom, CTO of StrataCloud, says IT leaders should take it one step at a time. He suggests following the tick-tock IT model – implementing a small infrastructure change, then building a few new capabilities on top – with clearly defined objectives in mind.

In the move to the software-defined data center (SDDC), a measured approach is critical. As our SDDC eBook notes, building out a small piece of infrastructure for a database service can be a safe first step. The process is well documented, and databases offer plenty of small, non-critical workloads to choose from. Stay away from databases that are tied to revenue or are customer facing, such as a database that your company’s eCommerce application queries to populate inventory details for online shoppers.

A workload-specific database service for a non-critical application will provide an ideal testing environment. Suppose the IT department at Netflix is considering software-defining a database service. A database that affects viewing of “House of Cards” would be a poor choice. If anything went wrong, it could directly and immediately affect paying customers. But software-defining a database that tracks a smaller aspect of a less popular show – viewership of “Gilligan’s Island,” for example – would be safer to test.

Another good option is a project that’s still in development, because it won’t yet be tied to revenue or be visible to employees or customers. For example, if a developer is working on a new report for accounting, IT could software-define one of the databases the report queries to populate its results.

There are several tools available to assist with configuration and provisioning the new infrastructure and database service. Vendor relationships will dictate what tools IT can purchase and from whom. But IT leaders also need to ensure members of their team can understand and apply automation and orchestration.Finally, before launching the database service, it’s critical to clearly define performance goals for both the new infrastructure and the database service that runs on it.

Testing the System and Measuring Success

There are several ways to investigate how well the software-defined system performs on the new infrastructure. Start by asking the following questions:

  • How quickly was it deployed?
  • How quickly could data be moved from one physical storage system to another? Was the process seamless?
  • Was it possible to add storage capacity non-disruptively?
  • How does performance of the new database service compare to established service-level objectives?

A small-scale project can also provide insights into staffing and budget issues that can arise as you move to a more comprehensive implementation. As you work through your first SDDC project, consider the following:

  • Do you have adequate internal expertise for handling systems automation and orchestration?
  • If not, are budgets sufficient to cover training for your existing staff and support contracts with vendors?

If the project was successful, you’ll know you’re ready to begin a new tick-tock cycle: building out another new piece of infrastructure and testing new functionality on top. As you progress, you can move toward software-defining more visible applications, and toward streamlining critical business processes.

For more details on designing an effective SDDC strategy, download Stratacloud’s ebook, The Software Data Center Defined: Beyond the Hype and the Hope: Four Practical Steps to Get Started.

About the Author

Aaron Bawcom

Aaron Bawcom is CTO at StrataCloud.