Migrating VMware Infrastructure to AWS: Know How the ‘Puzzle’ Pieces Fit Together First

Andy Smith is VMware Certified Design Expert for Transitional Data Services and is responsible for designing and delivering enterprise data center modernization solutions.

The cloud is often a no-brainer for ground-up architectures, but migrating any kind of legacy infrastructure to a cloud provider can be a very different matter. That's because you’re often  forced to address constraints that you might not otherwise have to deal with in your own environment. So much so, that if not approached the right way, companies often find that their cloud migration isn’t actually delivering any advantage. VMware environments provide an excellent case in point. By now most companies leverage VMware architecture, and for companies that are heavily invested in it, moving to the cloud has been somewhat problematic, often requiring a complete redesign to conform with the provider.

For example, when you migrate VMware to the cloud, you must align the formats of the virtual machine to the provider’s format. And typically moving to the cloud means going to a highly constrained data center, particularly in terms of the size of virtual machine that they have to offer. You have to map your services over, perform load balancing in accordance with what the provider offers, and possibly change IP addresses. These are, of course, just a few of the complicating factors that the largest cloud provider, AWS, has made an earnest attempt to mitigate.

AWS/VMWare Partnership Lowers, Doesn't Eliminate the Bar

When VMware and AWS announced their partnership, the news was greeted with appropriate fanfare. Essentially you could retain your virtual architecture, maintaining all the features of VMware you rely on but gain the benefits of residing in AWS, with Amazon maintaining the bare metal and network infrastructure. By leveraging what is basically a VMware environment that happens to reside on AWS, you can move workloads back and forth without bringing your system down and maintain security and other important configurations. By standing up a virtual data center in AWS as an extension to an existing physical data center, you could easily migrate application workloads to a VMware Cloud on AWS without significantly changing your architecture. It allows someone with a VMware environment to leverage the scaling, capacity, and burstability of being able to put servers in AWS while preserving many years’ worth of work. In theory, at least, there are still a few other very important considerations.

But it’s only half the battle

In reality, even with the ability to run VMware natively on AWS, you can’t just move an application over without accounting for all its dependencies, which includes VM-to-VM dependencies as well as multiple external application dependencies. A VMware environment can be viewed as a puzzle, where if all the pieces aren’t in proper place, you’re going to experience major challenges, even if you're moving into a cloud environment that natively supports VMware. During migrations, one of the biggest challenges is ensuring that critical applications remain resilient and continue to operate with no unplanned outages. For instance, if you’ve identified a Tier 1, mission critical app, but you haven’t accounted for relationships (dependencies) that it shares with Tier 1 or 3 apps, then you will likely break the Tier 1 applications by ignoring these dependencies. All too often, this is the kind of thing that companies find out on migration day or during a recovery event, at which point a critical app ceases to function; and you now must deal with an unplanned outage.  

Preventing this requires a complete understanding of both the architectures and dependencies of the applications. Modern IT infrastructure is a complex web of highly distributed, interconnected applications working in concert and running on a mixture of shared and often siloed conventional and virtual hardware, as well as cloud services. As Dell EMC CTO John Roese recently opined, the average enterprise’s infrastructure may even be more complicated than hyperscale providers like Google and Microsoft.

Completing the Puzzle

So how do you complete the puzzle? Of course, you start by leveraging all of the discovery tools at your disposal. While there are tools that attempt to rationalize where workloads should live based on cost and performance--or the machine logic--in order to avoid major problems and minimize downtime, you have to understand the business logic. You need to combine information from available sources, including DCIMs and CMDBs, and augment this with other available sources to create a unified, actionable view of your environment. This includes an aggregation of business factors such as resiliency, compliance and performance targets. If you don’t understand the business impact, you can’t judge the criticality of the application and infrastructure.

So, on top of all this, you’ve must have a more highly abstracted layer of understanding that includes the knowledge that can be deciphered by machines and the interdependencies that reside in departmental data silos, and in the brains of your IT department staffers, as well as other departments (or "shadow IT”, if you will). Ultimately, it’s the business logic, combined with the machine logic, that provides the complete picture of how applications in a VMware environment connect together as pieces to the same puzzle.

In summary, AWS and VMware have reduced the barrier for entry but not necessarily eliminated it. Understanding the dependencies between the applications running in a VMware environment is a necessary step, without which you’re essentially faced with having to re-work a puzzle in a new environment. There are many ways to migrate, and most will fail. The key is knowing the full application environment well enough to account for blocking dependencies before you take action, so that you’re not surprised during a migration or recovery event. To avoid a chaotic VMware migration, it’s best to start with a full understanding of how the puzzle pieces connect in your current environment first.

Opinions expressed in the article above do not necessarily reflect the opinions of Data Center Knowledge and Informa.

Industry Perspectives is a content channel at Data Center Knowledge highlighting thought leadership in the data center arena. See our guidelines and submission process for information on participating.