Stop treating AWS like another data centre

An enterprise in the financial services sector expected to save money when it migrated a major application to the cloud. Instead, a proposed ‘lift-and-shift’ strategy risked accruing an annual bill of $4.5 million. This type of revelation has made many businesses question the value of using hyperscale cloud infrastructure.

But a different approach slashed that bill down to $300 000. What happened? It started with addressing a crucial misunderstanding: the public cloud isn’t simply another data centre, and cloud savings materialise when focusing on an application’s end state and business relevance.

Cloud costs continue to confound companies, pushing many to radical steps such as repatriation (48% of respondents in a 2025 Cloud Computing Survey from Foundry said they are repatriating because of costs). Others simply absorb the financial blow, rationalising benefits in other areas.

Most made the same mistake, burning through budgets because they treat public clouds like AWS as another data centre, says Jonathan Leibbrandt, Senior Solutions Architect at TrueMark.

“When people move to the cloud, they want to provision it like for like, as they have it in a data centre. If they’ve got a virtual machine running in their data centre, they say, ‘We need to get the exact same virtual machine, same spec and everything running in AWS.’ But that’s not the best approach.”

A company moves applications to a public cloud provider and retires the original servers. To avoid complications, they replicate the original server’s hardware specifications. But this strategy overlooks the most fundamental difference between bought servers and cloud infrastructure.

“In data centres, you’re buying tin that you need to make last for several years. You’re thinking about peak capacity, future growth and what that application might need before your next hardware refresh. Once you’ve bought that capacity, you’ve bought it. If the server is sitting at 10% utilisation for most of the day, you’ve already paid for it. But we’ve moved from buying capacity to consuming it. You don’t have the same three-to-five-year hardware cycle anymore. The commercial model has changed.”

There are reasons why this procurement mindset persists even in a consumption model. Moving a service like for like often seems simpler. Companies avoid tinkering with established and customised applications, because refactoring and rearchitecting can seem risky and costly. Migration projects dealing with hundreds of servers and cutovers often reinforce like-for-like approaches.

But these choices create costs. Leibbrandt explains how that happens. A company decides to move 2 000 virtual machines to AWS, spinning up 2 000 EC2 instances and running the applications in the same way. But since this strategy uses capital expenditure logic, consumption quickly erodes the margins.

“If you have 2 000 servers running 24/7, you’re paying for something that you’re probably not even using 24/7.” 

What is the alternative? In this example, Leibbrandt recommends moving to an event-driven architecture that focuses on serverless technologies. Infrastructure spins up to complete a transaction then spins down again, minimising idle costs.

“You’re only paying for those couple of milliseconds that it takes to complete that transaction.”

Hence the trap of treating public clouds like AWS as another data centre, because it’s not. It’s a different operating model to the predict-buy-sweat procurement cycle, requiring a different approach. Leibbrandt says companies should do the following when deciding on a migration:

• Engage with a migration partner: “We don’t charge to do discovery with a customer. We’ll have a conversation and do an assessment, and create a business case.”
• Develop a migration strategy: “Understand what exists today, go through applications, go through source code and create a migration plan to decide if AWS is the right option or not.”
• Establish internal skills: “People inside the business with the right skills can connect what the business has and what the cloud can provide. This is why we work very closely with our customers to upskill. The customer shouldn’t end up with a black box.”
• Be prepared to adapt: “I’m not against lift-and-shift or think everything belongs in the cloud. But there are more ways to do things than a virtual machine; finding those ways are where the biggest savings and improvements happen.”

Most companies treat the public cloud as if it’s another data centre, and end up paying much more. They really shouldn’t, and if they adopt the right approach from the start, they won’t. They’ll see real savings and an investment that genuinely grows with them.

Duplicating a server is tempting. But is it the best choice? That’s the question to answer when migrating to the public cloud.