Assess before moving anything
What exists, what is actually used and what is worth moving, rebuilding or switching off. Lifting an underused server as-is is the most expensive way to be in the cloud.
Moving to the cloud is easy; making it pay off and still standing a year later is not. We design, migrate and leave the operation documented, with spend under control from the first month.
What exists, what is actually used and what is worth moving, rebuilding or switching off. Lifting an underused server as-is is the most expensive way to be in the cloud.
The least painful things first, so we learn with a safety net. With a rollback window on every wave and no heroic weekends.
Entra ID, multi-factor, conditional access, and the principle that everyone reaches only as far as they need. Most incidents don't come through a technical flaw: they come through a credential.
A backup that has never been restored isn't a backup, it's a hope. We define recovery objectives and run the restore test with you watching.
Tags by department, budgets with alerts, shutting down what isn't used outside working hours, and a monthly review. An Azure bill grows on its own if nobody looks at it.
Everything below is in real projects. If something isn't on the list, we tell you before we start and not after.
Not always, and anyone who says yes without looking at your case is selling. It pays off when the load varies, when the hardware is due for renewal, or when the real cost of keeping it in-house includes someone maintaining it. We run the numbers on both scenarios before deciding.
The target is zero, achieved by migrating in waves with prior synchronisation. Systems that can't be replicated have a window, agreed in advance and with a written rollback plan.
It does, like everything. The right question is what happens then: which zone it is replicated to, how long it takes to bring back up and how much data is lost in the worst case. Those three numbers are agreed at design time, not when it happens.
Partly, as with any cloud. It's mitigated by using containers and standard databases instead of proprietary services where that is reasonable. We flag it when a decision increases the dependency, so it is a choice and not a surprise.
Yes, with an hour bank or a service agreement. But the documentation is handed over either way: the idea is that you can change provider without going blind.
Twenty minutes is enough to know whether there's a project. If there isn't, we'll say so.