Finance proposes a three-year commitment covering 80% of current compute. Engineering is planning a container migration. Advise.

What the interviewer is testing

Whether you recognise that the risk of a commitment is architectural rather than financial.

The conflict

A three-year commitment to a specific instance family assumes you will still be running that shape in three years. The container migration changes the compute profile — different instance types, higher density, possibly a managed service or serverless.

The failure mode is worse than the wasted money: the commitment becomes an argument against a good architectural decision. "We can't move to containers, we've pre-paid for these instances" is a conversation that happens, and it is the worst possible outcome.

The advice

Commit only to the stable baseline — the floor of usage that will persist regardless of architectural direction. If the migration is real, 80% is too high; the right number is what will still be running after it.

Prefer flexible commitment forms that apply across instance families and regions over rigid ones tied to a specific configuration. The discount is slightly smaller and the optionality is worth more than the difference, particularly during a migration.

Ladder the expiry dates rather than committing everything at once, so the estate is never fully locked in a single window.

Sequence it: commit for one year now covering the baseline, complete the migration, then commit for three years against the new steady-state profile. That captures most of the discount without betting on an architecture you are actively changing.

The organisational point

This decision needs finance and engineering in the same conversation, because the commitment horizon must be set against the architectural roadmap and neither function has both pieces of information on its own. Finance sees usage; engineering sees the plan.

What a strong answer adds

Modelling it: the discount foregone by committing to 50% instead of 80% for one year, against the cost of stranded commitment if the migration proceeds. The numbers usually make the case clearly, and presenting them turns a disagreement into a decision.

Common weak answers

Rejecting commitment entirely, which leaves a large discount unclaimed. Accepting the proposal and hoping the migration slips.