How businesses can avoid and navigate cloud vendor lock in

I’ve lost count of how many times I’ve seen ambitious businesses rush into hyperscale cloud agreements. The pattern is nearly always the same: they’re keen to get going, are naively blinded by the lure of free credits, and tempted by the prospect of instant server spin-up. It’s not surprising then that these companies often feel a quick signature on the agreement will instantly sort most of their infrastructure headaches.

For the first few months, this can be the case. Workloads scale effortlessly, APIs integrate seamlessly, and their hyperscale cloud setup runs like a dream. But when the last of the free credits run out, the reality becomes clear. They’re not only utilising a service from a hyperscaler, their entire infrastructure – and company – has been built around it.

What’s left is a business so heavily entwined with their cloud provider that they’re dependent on it. Every system and process is tied to one provider, making getting out feel nigh on impossible.

Whether it be a startup or a high growth scale up, how can others avoid cloud vendor lock in?

The risks

In my experience, there are two main risks that come with hyperscale cloud. The most significant is technical entrenchment. With so many proprietary products and services on offer, like application platforms and managed databases, the temptation for a business to buy it all is understandable, particularly when financial sweeteners are on the table.

The frictionless nature of these tools is a big part of their appeal. The fact that they work seamlessly with one another, and their native ecosystem, is very appealing. But it also means the portability of this infrastructure drastically decreases as time goes on. More tools, more inter-dependent services, and a web of capabilities that only work in their home environment means migration quickly becomes a game of total re-engineering.

The second problem with hyperscale environments is that costs can escalate, and fast. Workloads accumulate and usage grows, yet hyperscale cloud is often not optimised to run them in the most efficient way. With variable charges and a lack of transparency around billing, predicting spend can become difficult. It’s not an uncommon problem, either – 62% of businesses exceeded their cloud storage budgets last year.

Why scaleups are vulnerable

It’s easy to see how the hyperscale model makes entry smooth and appealing. Especially for early-stage companies, fuelled by a desire to ‘get going’ as quickly as possible, the prospect of being able to scale seamlessly with a hyperscaler seems like a no brainer. After all, this flexibility and a low barrier to entry means rapid business growth, higher turnovers, and outpacing the competition faster, right? It may seem appealing at first, but it’s a slippery slope to hyperscale cloud dependency.

In the early months of an agreement, these scale up companies will quickly onboard their provider’s proprietary tools – all of which require little engineering. Technical leaders are quick to say that optimising their infrastructure is a “tomorrow” problem. The focus is rapid deployment and growth, throwing caution to the wind when it comes to the infrastructure back end. 

Soon enough the business has burned through its free credits and is deeply embedded within its provider’s ecosystem. Their tooling, processes, and beyond have become completely dependent on that single environment. Moving away from this provider suddenly becomes an incredibly expensive and painstaking task, and for many scale ups, they simply do not have the cash or resources to make this feasible. The reality soon sets in that they’ve traded future flexibility for short-term convenience.

Locked-in? Here’s what’s next…

Despite the challenges, there is some hope for businesses at this point of lock in. The first step is establishing a firm grasp on where these dependencies are, meaning a full audit of technical services and tools is needed. With data deeply entwined in these areas, this is often the most complex and costly aspect of any cloud repatriation. 

But before any data or systems are pulled away, it’s worth stepping back and considering the real scaling needs of the business. Are there aspects of your business operations that could comfortably run on bare metal, for example? How often does customer or business demand require the near-instant bursts of compute capability that the hyperscalers often support? It’s easy to over-consider the ‘what ifs’ and opt in for more hyperscale cloud capacity than needed, just in case. So, properly considering your predictable and unpredictable workloads is key.

The reality is that most businesses find they sit somewhere in the middle, needing the instant scalability of cloud combined with optimised compute power provided by bare metal servers. A hybrid approach is a practical option, pulling stable workloads out of a hyperscale environment and keeping the elasticity of cloud in place for when it is needed. 

Truthfully, the best cloud repatriation projects I’ve seen focus on an incremental move away from hyperscale environments, migrating data on a step-by-step basis to carefully balance costs and operational stability. The key to achieving future flexibility in an optimised hybrid bare metal environment is to not cut all strings overnight.

Re-setting the relationship with cloud

For the countless businesses I’ve advised over the years, cloud itself isn’t the enemy. In fact, cloud can be an ally – a perfect companion in hybrid environments where quick scalability and flexibility is needed alongside raw compute power. The true danger lies in speeding ahead and going all-in with a hyperscale cloud set up without a thought to future infrastructure needs. It may be easy to set up, but getting out will always be challenging.