4 Mistakes that Derail Cloud Builds

Public cloud platforms such as Microsoft Azure make it incredibly easy to spin up new computing instances. But that doesn’t mean the process of architecting new cloud environments and migrating data is always simple.

When organizations aren’t careful, they can end up with messy cloud environments that lack a foundational structure, where even basic features fail to function as expected. Occasionally, we at CDW get called in to help customers clean up such environments, which are often the result of working with inexperienced partners who underestimate the work and knowledge needed to complete a large cloud migration.

Watch out for these four common mistakes that can quickly derail business-critical cloud builds.

Thorough planning should be table stakes for any cloud initiative, but, unfortunately, some partners skip this foundational step. We recently worked to help a customer recover from a subpar engagement with a partner that essentially slapped some resources together and expected them to work without even the basic hub-and-spoke architecture necessary to optimize the workloads. If a partner can’t explain the basic architecture of a cloud environment, as well as the reasoning behind that blueprint, run.

Beyond a basic blueprint, cloud deployments require organizations and their IT partners to make thoughtful decisions about how to segment workloads, clarify network paths, and create logical boundaries that support operational control and business continuity. When architecture isn’t optimized for an organization’s workloads and business needs, significant problems can emerge. For instance, if workloads aren’t properly isolated, a failure in one virtual machine can bring down several others — an issue that can easily be prevented by making the right architecture decisions from the start.

We’ve seen customers with misconfigurations in their cloud environments. The real danger lies not so much in a single misconfigured virtual machine but in the way cloud infrastructure is tied together. When organizations and their partners don’t consider dependencies between different components of their cloud environments, seemingly small issues can have huge ripple effects. One customer we worked with found that his organization’s virtual machines would occasionally stop working without warning. When the affected VMs were rebooted, two other applications would go down. The dependencies were so tangled that working on one component of the cloud environment would inevitably affect some other element of the environment.

Sometimes, the worst thing a cloud partner can do is try to fix problems when they don’t understand the underlying cause. If a partner lacks the knowledge to set up a cloud environment properly in the first place, they probably also lack the skills to effectively troubleshoot when issues arise. Occasionally, when customers have called us in to remediate cloud environments that were poorly designed, we’ve had to simply stand up a parallel environment from scratch. When something is fatally flawed, sometimes the only option is to start over.