5 min read

Cloud migration without the 2am page: a pre-flight checklist for Philippine SMBs

Six things to lock down before a weekend cutover — credentials, data, rollback, monitoring, comms, and the human handoff.

  • Cloud Migration
  • AWS
  • Azure
  • GCP

Most weekend cutover disasters are not infrastructure failures. They're missing paperwork — un-rotated credentials, no rollback runbook, missing observability, or no one on call. Here's the six-item checklist we run through with every client before a Saturday-night cutover.

1. Credentials and access

  • New cloud account MFA enforced; no long-lived access keys on production workloads
  • Old provider credentials rotated AFTER cutover, not before
  • Least-privilege IAM roles in writing — who can do what, audited
  • Break-glass admin credentials stored in a password manager, not in chat

2. Data migration verification

Don't trust 'the sync finished.' Verify with row counts, checksums on critical tables, and a parallel read window before cutover.

  • Row counts match (per table, per archive)
  • Checksums on blobs/object storage
  • Sample-record integrity check for 50+ records across critical tables
  • Both environments reading for at least 1 hour before cutover (shadow read)

3. Rollback runbook (in writing)

Every cutover must have an exit criterion: if X happens by 23:00, we roll back. Decide this BEFORE Saturday, not at 23:00.

4. Monitoring and alerting

  • Both old and new environments monitored (not just the new one)
  • P1 alert channels: who gets paged, how, with what runbook link
  • Synthetic check that exercises the user-facing critical path (login → checkout → confirmation)
  • Cost alarm set on day-1 budget (you don't want a runaway loop draining the account)

5. Communications plan

Who tells your customers, when, and how? The best cutovers announce a maintenance window in advance; the worst leak via 'why is the site down?' tweets at 22:30.

6. The human handoff

  • Documentation delivered as a runbook (not a PDF nobody reads)
  • Two handover sessions: pre-cutover walkthrough + post-cutover review
  • 30 days of included on-call coverage for the new environment
  • Named contact for any issue in that 30-day window

How we run cutovers

428AM does weekend cutovers for Philippine SMBs moving from on-prem or shared hosting to AWS, Azure, or GCP. Cost-aware design (we scope to your workload, not to the cloud provider's upsell), 30 days on-call after, written runbooks your team actually reads. Timeline for a typical cutover: 4-10 weeks of prep, one weekend of execution.

Working on something this touches? Send us the scope — we'll reply within 1 business day (PHT).

Chat on Messenger