Cutover weekend gets all the attention, but the month after cutover decides whether the migration was actually good. This is the month-two checklist: five jobs, none glamorous, that separate the businesses happy with their migration from the ones calling someone like us a year later.
1. Decommission the old gear, on a date
Once the new environment has survived two real weeks, the old server's retirement gets a calendar date. The sequence: confirm nothing still points at it (watch its network traffic for a week; the forgotten label-printer service will announce itself), take a final full image as a keepsake backup, wipe the disks properly, then power down. The alternative timeline, where it hums unpatched in the closet for two years, is mistake number six.
2. Start the cost habit now
Month two is when the first real bills arrive, and when the review habit either forms or doesn't. Twenty minutes monthly: compare against last month, name anything new, delete orphaned resources, and downsize anything idling under 20 percent. Quarterly, do the fuller right-sizing and reserved-pricing pass. Bills reviewed from month two stay flat. Bills first reviewed in month eighteen fund our rescue practice. The mechanics are in what a migration really costs.
3. Wire up monitoring and alerts
The closet server got glanced at every time someone walked past. Nobody walks past a cloud console. Set up uptime checks and alerts for the things that matter: the app is answering, the VM isn't at 95 percent disk, the backup job ran, the bill didn't spike. Alerts go to a human who acts, and the escalation path gets written down. "The cloud is reliable" is true of the platform and says nothing about your configuration on it.
4. Review access while it's fresh
Migrations scatter permissions: the consultant's admin account, the vendor's temporary access, the broad everyone-can-edit grant from a hurried Tuesday. Month two is the audit moment, while everyone still remembers why each grant exists. Remove what's stale, narrow what's broad, confirm MFA on everything administrative, and re-check against least privilege. Put the same review on a twice-yearly calendar, because permissions only ever accumulate.
5. Back up the cloud itself
The provider's redundancy protects against their disk failing, not against your ransomware, your accidental deletion, or your compromised admin account. Microsoft 365 and Google Workspace want independent backup; cloud VMs want snapshot schedules plus offsite copies; and the whole arrangement wants one restore drill before you declare victory. Then update the continuity plan, because half its pages just changed: new systems list, new recovery paths, new vendor numbers.
Then stop
Resist the month-three temptation to immediately optimize into architectures you don't need. Let the new normal settle for a quarter, collect real usage data, and revisit. The migration is done when the checklist is done and nobody talks about the server anymore. That silence is what success sounds like.
Want this handled instead of homeworked? That's the job.
Email us →