Lift and shift, and the regrets thereof

Lift and shift promises the one thing steering committees want to hear: the move without the change. Take the VMs as they are, land them on someone else's hypervisor, done by Q3. Sometimes that is exactly right. Mostly it buys the same estate with new invoices, and the regrets arrive in a fixed order.

Regret one: the bill. On-prem estates are sized for peak plus fear: that 16-core box running at 6% was paid for years ago, and idling is free. The cloud meters the same fear monthly. Lifted estates routinely land 30-50% over the business case, because the business case priced the servers and the cloud prices the habits. Rightsizing was the actual project, and lifting deferred it to a place with a running meter.

Regret two: the estate did not get younger. The 2016 boxes are still 2016 boxes, just in Virginia now, with the same end-of-support clock ticking. The patching debt, the snowflake configs, the unknown batch layer: all faithfully preserved. Lift and shift is a moving company, not a renovation. Nobody should be surprised, and everybody is.

Regret three: the exit narrowed. The lifted estate quietly grows roots (the load balancer becomes theirs, the backup becomes their service, the identity plane binds) and the "we can always leave" clause becomes theoretical. The repatriation crowd has receipts on what leaving costs at petabyte scale; even with egress fees waived, it is a logistics project, not a decision memo.

And yet. Lifting wins in three narrow cases, and we recommend it there without apology: the datacenter has a closing date and there is no time for surgery; the workload is a stable appliance you will retire within two years (park it, then kill it, and skip the renovation entirely); or the org needs one consolidated substrate before the real work, and knows it is buying a staging area, not a destination.

The failure mode is not the lift. It is the sentence "we will optimize after we land," spoken by organizations that have never once optimized after landing, because the pressure died the day the old building emptied. If the plan says modernize-after-move, put the modernization on the roadmap with dates and owners before the trucks roll, or write the honest version instead: we are moving the problem, on purpose, and here is what that costs per month to keep. Both are defensible plans. Only one of them gets written down.