Control / access
Status: untested
The cloud was never yours.
Cloud outages, vendor lock-in and geopolitical risk are turning exit plans into infrastructure requirements. See what fails, what control costs, and when modern on-prem makes sense.
Exit readiness
Can you leave—or only log in?
If the answer to any of these is “we think so,” you have a dependency worth testing.
Export
Can you extract your data in a usable format, within a known time and cost?
Identity
Can recovery work if the primary provider account or login is unavailable?
Backup
Are your backups outside the same account, control plane and deletion authority?
Replace
Which managed APIs, databases and observability systems must be rebuilt elsewhere?
Rehearse
Has the exit run successfully—or does it exist only in a diagram?
Incident dossiers
The problem already exists.
The provocative claim comes first. The provider’s own postmortem comes next.
Your redundancy has a blast radius.
AWS, Azure and Cloudflare postmortems show how shared control planes turn small defects into broad failures.
Open dossier →Your data can survive while your access dies.
Geographic copies do not protect against account-level deletion, suspension or a shared administrator.
Open dossier →Your region is not your control boundary.
A region says where workloads run—not who owns, governs or can interrupt the service. Europe provides a documented example.
Open dossier →Rent is simple. Until it is permanent.
Stable workloads deserve a real comparison: cloud premium versus hardware, colocation, people and an exit you can test.
Open dossier →Going back can be moving forward.
Cloud repatriation is workload placement with new evidence—not nostalgia for a server closet.
Open dossier →Cloud is excellent at elasticity.
- Fast experiments
- Demand that changes quickly
- Managed capabilities worth their premium
It is not ownership.
- Stable, highly utilized workloads
- Critical data and offline continuity
- Systems that need a real exit path
Modern on-prem is not a rack humming under someone’s desk. It can be owned hardware in professional colocation, a private cloud, or a hybrid system designed around your control boundary.
It is not the answer to every workload. It is an option worth making real before you need it.
Build an exit plan →