Do I need a service mesh for zero-downtime?

No. A mesh (Istio/Linkerd) makes mirroring and weighted routing easier, but you can do zero-downtime with NGINX/Envoy, DNS weights (Route 53), and good discipline. The key is progressive traffic shifting and automated SLO gates.

How do I handle schema changes without downtime?

Use online schema tools (`gh-ost`, `pt-online-schema-change`) and versioned contracts. Apply backward-compatible changes first (expand), run dual-writes, backfill, then switch reads, then contract.

What’s the minimum I need to automate?

Automate rollbacks, traffic weight changes, and SLO checks. Humans can watch dashboards, but the brakes must be automatic. If you can’t test the rollback in staging, you don’t have it.

What about cost and performance during the window?

Expect 1.5–2× capacity for an hour to absorb mirroring and canary overhead. Plan it, get sign-off, and right-size after the hold period.

Guides · Dec 1, 2025 · 9 minute read

The Zero-Downtime Cutover Checklist We Use When Failure Isn’t an Option

A field-tested, step-by-step plan to move a critical workload without users noticing—and without praying to the DNS gods.

Back to all posts

The Zero-Downtime Cutover Checklist We Use When Failure Isn’t an Option

Key takeaways

Implementation checklist