Cloud repatriation spotlights the need to avoid letting hype determine your analytics strategy. Paige Roberts, Open Source relations manager at Vertica, gets to the bottom of what reverse cloud migration is all about and the purpose it may serve.
For the past decade or so, businesses have been migrating to the cloud in droves. No surprises there, as cloud technology has been nothing short of revolutionary, bringing forth unimaginable opportunities even for the smallest of businesses. That said, not every cloud migration results in an outcome that is expected. Yes, it can be easy to get everything up and running; it can be easy to support and maintain; resources can be spun up on demand. But as businesses start to scale, some realize the cloud model isn’t working for them.
Three Main Reasons for Cloud Repatriation
Cloud repatriation or “reverse cloud migration” is a process of pulling back an organization’s applications, data, and workloads from a public cloud and putting them back onto a local infrastructure, such as an on site or co located data center or a private cloud. Studies show that 54%Opens a new window of organizations in 2024 repatriated some or all of their workloads or moved data away from the public cloud. Here are three main reasons why:
1. Cost, transparency and predictability
The idea of paying just a fraction of the cost upfront and moving away from a cap ex model to an op ex one is really hard to resist. This is why most organizations migrated to the cloud in the first place. Unfortunately, organizations are now realizing that the cloud isn’t as financially appealing as it once appeared to be. Cloud pricing is based on a pay as you go model the higher your utilization, the greater your cloud bills. What’s more, studies show over 90% ofOpens a new window enterprises overshoot their cloud budgets because they either underutilize or overprovision resources or lack the ability to properly manage cloud environments.
Cloud providers also have non standard pricing, which means that every cloud provider will have different cost definitions (usage limits, tiers, etc.) and charges will vary greatly based on services used (security and management of tools, database writes, etc.). There are also a number of hidden charges involved. For example, moving your data out of the cloud provider can attract egress fees.
Cloud providers can lock customers in long contracts or tweak their pricing suddenly, which can cause cloud bills to skyrocket. In contrast, on premises or co location offers simpler and more predictable pricing. You buy a server that costs X, and you pay your co location fee Y per month and that pretty much sums it up. The hype machine makes two big statements about cloud computing, claiming it will save costs and require far less expensive skillsets to manage. The evidence counters both those statements.
2. Cloud performance
Although the cloud provides theoretically limitless scalability, it still loses some speed due to internet connectivity and virtualization overhead. In most cases, cloud performance issues like noisy neighbors in multi tenant software don’t matter because it’s plenty fast enough to serve the use case. However, for certain use cases or for higher scales of data, workloads, or concurrency requirements, faster performance is essential.
Some real time analytics workloads, like machine learning based AI, can be sensitive to latency. Analytics applications can leverage caching and other network optimization methods to reduce latency. However, one of the easiest and most pragmatic fixes is to shorten the communication path. This means that unless the data was born on the cloud, bring the analytics back in house.
3. Security, control and regulatory concerns
As long as security is set up correctly and best practices are followed, there is no reason why a public cloud would be any less secure than an on premises environment. That said, one can never rule out the possibility of cloud misconfigurations, which are the number one cause of cloud data breaches globally. Even leading cloud providers like Amazon and Microsoft have been victims of leaks and breaches due to cloud misconfiguration issues.
Regulations regarding the geographical location or the sensitivity of data can be a major concern, requiring data processing outside a public cloud or requiring data to remain geographically contained. Businesses may find it easier to relocate that data back in house, even in less regulated regions, to avoid potential problems from future regulatory changes. Transferring hundreds of terabytes, petabytes, or even exabytes of data from one location to another can be very expensive, and on premises is the one location no regulation is likely to rule out.