Quick answer: Planned maintenance is scheduled work on the network, power or host hardware, announced in advance where possible and placed in low-traffic hours when practical. Announced windows are not outages and are excluded from SLA incident handling. Exact notice practices are governed by the Terms of Service.
Overview
Proactive maintenance is what keeps infrastructure from failing unpredictably: firmware and security updates on network gear, hypervisor patches, capacity upgrades, UPS and generator tests. The policy trade-off is deliberate — short, announced, controlled windows instead of long, surprise incidents.
Before you start
- Read maintenance announcements when they arrive (see how notifications work) and diarise relevant windows.
- Know your service's location so you can filter announcements to what actually affects you.
- For critical systems, design for graceful degradation: caches, queues and retry logic turn a blip into a non-event.
Step-by-step guide
How to live with planned windows painlessly:
- On receiving an announcement, check scope and time against your services.
- Reschedule deployments, migrations, backups and cron-heavy jobs out of the window.
- If expected impact is stated as connectivity interruptions, inform your own customers pre-emptively.
- During the window, resist restarting/reconfiguring things — transient errors are expected and self-heal.
- After the window, verify your stack came back cleanly (services up, replication caught up, queues drained).
Common issues
- Monitoring alarms during the window: annotate or silence checks for announced windows to keep alert fatigue down.
- Long-lived connections dropped: network maintenance can reset sessions — apps should reconnect automatically.
- Assuming zero impact: "no expected impact" is the plan, not a guarantee — critical operations are still better scheduled elsewhere.
When to contact support
Open a ticket if a planned window would collide with a critical business moment (support can sometimes advise on timing), or if problems persist after the announced window closes.
Frequently asked questions
Does planned maintenance count as downtime under the SLA?
Announced maintenance windows are excluded from SLA incident handling. They are scheduled, communicated in advance where possible, and placed in low-traffic hours when practical.
What should I avoid doing during a maintenance window?
Deployments, migrations, large backups and invasive troubleshooting. Transient errors inside the window are expected; verify your stack after the window closes instead.
Can a maintenance window be moved for my critical event?
Sometimes scope allows flexibility. Tell support about the conflict via ticket as early as possible and the team will advise what is realistic for that specific work.
Related articles
Need a hand? Contact Cloud2Y support →
