Pythian Blog: Technical Track

Are reverse cloud migrations a thing?

At Pythian, we spend the vast majority of our time helping our customers take full advantage of the public cloud. But there is always that tiny percentage that comes to us asking if we can move them from the cloud back to on-premises. Whenever a customer brings up a “reverse migration”, we always ask what’s driving the request and, over time, we have seen some common themes.

Common challenges with public cloud implementations:

  1. Costs are higher than expected.
  2. Performance is not as expected.
  3. Management of cloud estates is more challenging than expected, especially in hybrid architectures.
It’s not that they aren’t seeing lower costs or improved performance or a reduction of management effort, it's that expectations were out of line with reality. It's also interesting that security concerns haven’t yet come up as a driver for reverse migrations - likely because cloud providers are doing amazing work here and most of our clients have strong DevSecOps built into their managed services with us. Inevitably, after we uncover what’s behind their request to move out of the cloud, we end up developing a plan of action that keeps them happy and realizing more value in their cloud world.

Three ways to avoid the need for a reverse migration

Even if you aren’t considering a move out of the public cloud, there are some best practices that you should focus on if you want to optimize for lower costs, improved performance and ongoing management of your cloud estate. Our top three are:
  1. Invest in Cloud FinOps.
  2. Optimize your cloud performance.
  3. Get a grip on cloud governance and management
  1. Cloud FinOps: The ongoing cost management of the cloud was not top of mind when the cloud was first embraced. Everyone was focused on the obvious benefits of scalability, flexibility and new services and tools. After a few months of mounting cloud bills, CFOs started to get engaged and the conversation shifted - and rightly so. It’s easy to see that costs can grow dramatically. There is a myriad of reasons for this - anything from not taking advantage of reserved instances to architectures designed without cost in mind to cloud “sprawl” as services are turned on and left on even when not in use. Implementing a strong FinOps process can result in significant reductions in cloud costs.
  2. Improved performance: This is a goal of many of our clients and while it can often be realized in a cloud world, it’s not a given. Poorly architectured platforms are a key culprit as customers often apply traditional world thinking to their new cloud world and as a result, don’t see the benefits that they were expecting. Rearchitecting for the cloud is often the best way to address this concern, not moving back to on-prem.
  3. Cloud Managed Services: Cloud services, especially PaaS were heavily promoted as almost “hands-free” in terms of management, but in a world where it's dead simple to adopt new services on different clouds, our clients are seeing explosive growth in the number of services and frameworks being used by different dev teams, often on different clouds, leaving IT Ops with an ever-expanding management problem. While migrating back to on-premises might allow them to get control back, a better solution is to improve cloud governance and/or outsource cloud management to experts to get everything back under control while still taking advantage of the good parts of the cloud.
What we are seeing is a desire to more easily move cloud-native workloads between private and public cloud infrastructure with proprietary hybrid cloud products such as AWS Outposts, Google Anthos and Azure Stack. This has become a reality. This can simplify management processes and makes migration between public and private environments more seamless, enabling enterprises to more freely place workloads as close as possible to the other IT systems and the users that interact with them. This geographic flexibility can potentially boost performance and even reduce costs as long as IT operations have the tools they need to manage across this shifting world. So, the bottom line is that well-architected systems with good cloud governance, coupled with flexible cloud operations and strong FinOps across multiple clouds and hybrid environments, is key to keeping the business and the technical arms of the company working hand in hand and extracting value from cloud without ever having to go back to the old world and all of its restrictions.

No Comments Yet

Let us know what you think

Subscribe by email