I have sat through a lot of cloud migration retrospectives, and the disappointed ones nearly always tell the same story. Leadership expected the cloud to make things cheaper, faster, and more reliable. Instead the bill went up, the reliability stayed the same, and nobody could quite explain why. When I dig in, the cause is usually that the team lifted and shifted: they took the on-premises servers exactly as they were and ran them, unchanged, on rented hardware.
Lift and shift is seductive because it is the least work up front. You do not rearchitect anything. You do not question any assumptions. You just move the virtual machines and point the DNS at the new place. And for a narrow goal, like getting out of a data center whose lease is ending, that is a perfectly reasonable tactic. The trap is expecting cloud benefits from it. You moved the mess to a more expensive address and kept every inefficiency that made it a mess.
The value in cloud is not the location. It is the operating model: elasticity, managed services, paying for what you use, and infrastructure defined in code you can rebuild. A server that ran at fifteen percent utilization in your basement runs at fifteen percent utilization in the cloud too, except now you are paying premium rates for the idle eighty-five percent, twenty-four hours a day, and someone is surprised when the invoice arrives.
I am not arguing you must rearchitect everything before you migrate. That is the opposite mistake, and it stalls migrations for years chasing purity. The pragmatic path is to lift and shift what you must to hit the deadline, then treat that as the starting line, not the finish. Once the workload is stable in its new home, you go back and earn the actual benefits: right-size the instances, replace the self-managed database you were babysitting with a managed one, put an autoscaling group where a fixed fleet used to sit.
The mistake is calling the migration done when the servers boot in the new place. That is when the real work starts. Otherwise you have signed up for a more expensive version of exactly what you had, and told everyone it was progress.
– Serguey Shinder
Top comments (0)