Rightsizing Your Way to the Cloud
Moving overprovisioned workloads to the cloud as-is can cost more than keeping them on-premises. In the report, "Rightsizing Your Way to the Cloud," Bain & Company draws on AWS Migration Evaluator data to show how rightsizing before migration can cut costs by as much as 30% to 60%. Read the report to avoid lift-and-shift overspending, then contact us for help building a rightsized AWS migration plan.
Why doesn’t a move to the public cloud always lower costs?
Many organizations move to the cloud using a direct “lift and shift” approach—essentially a like-for-like transfer of existing servers, storage, and applications. The problem is that this often migrates existing inefficiencies instead of fixing them.
Based on an analysis of more than 60,000 workloads by Bain & Company and AWS:
- Poorly planned, direct-match migrations can be 10%–15% more expensive than keeping workloads on-premise.
- When IT leaders were surveyed, the top disappointment was that total cost of ownership did not decline after moving to the cloud.
The main culprit is overprovisioning. About 84% of on-premise workloads are overprovisioned, meaning they run with more compute and storage capacity than they actually use. When you move these workloads as-is, you simply pay cloud rates for the same excess capacity.
Other common issues include:
- Zombie servers that are no longer needed but never decommissioned.
- Fragmented and duplicated workloads spread across machines and data centers.
- Untuned software licensing tied to core processor counts that no one has rationalized.
Without a deliberate rightsizing effort before and during migration, the cloud can end up being a more expensive place to run the same inefficient environment.
What is rightsizing and how much can it actually save?
Rightsizing is the process of aligning your compute, storage, and licensing footprint with what your workloads actually need—before and during migration to the cloud.
Instead of copying your current environment, you:
- Eliminate zombie servers and wind down unnecessary activities.
- Match instance sizes to real usage patterns rather than peak assumptions.
- Use burstable instances and small reserves of excess capacity to handle spikes instead of overbuilding for the worst case.
- Analyze storage patterns (average and peak usage, proximity to compute) and downsize where possible.
- Rationalize core counts to optimize software licensing fees tied to processors (for example, Microsoft SQL licenses).
When companies take this more disciplined approach, the economics change significantly:
- Bain & AWS data shows that rightsizing can cut migration-related costs by 30%–60%.
- In one large-scale example, a direct-match migration of 105,000 server instances and 20 TB of storage would have increased costs by 22%.
- The same environment, when rightsized, would have reduced costs by 36%.
- Morningstar, after rightsizing with AWS Migration Evaluator, found that the majority of its instances would cost about 30% less in the cloud.
Rightsizing is less about cutting for the sake of cutting and more about reimagining how capacity is provisioned, monitored, and paid for so that the cloud’s flexibility actually shows up in your P&L.
Which workloads benefit most from a rightsized cloud migration?
Not every workload behaves the same way in the cloud, so prioritization matters. The data suggests that some categories offer more upside when you rightsize them thoughtfully.
Workloads that typically benefit most include:
- Nonproduction workloads (test, development, staging):
These tend to have more volatile usage, so they benefit from the cloud’s flexibility. You can scale them up and down and reduce software licensing fees tied to core processor counts.
- Nonvirtualized workloads:
These often see greater savings because they haven’t yet captured the efficiency gains that virtualization provides. Moving them to a rightsized cloud environment can be a step-change in how they’re provisioned and managed.
- Workloads with transferable licenses (for example, Microsoft SQL):
Organizations that can bring existing relational database licenses into the cloud often achieve the highest savings, especially when they also reduce core counts through rightsizing.
In practice, a pragmatic sequence looks like this:
- Start with nonproduction and noncritical workloads to build your rightsizing playbook.
- Target nonvirtualized environments where the gap to current best practices is largest.
- Layer in license-aware optimization, focusing on databases and software tied to processor counts.
This approach helps you capture early savings, refine your operating model, and then extend a rightsized strategy to more complex production workloads and hybrid environments.