The problem with maintenance windows is not only when the change is made, but whether it is prepared so that it does not create a second problem. A bad change at a good hour is still a bad change.
That is why we see maintenance as a controlled operation, not just as "a time when things may be touched".
What must be clear in advance
- what exactly is changing and why
- what the rollback path looks like
- which component is the risky centre of the change
- how you will quickly tell whether the result is good or bad
Related topics
Want to see these settings in action?
All our VPS services are built around the same engineering approach. If you are looking for real infrastructure rather than just a reseller service, see the VPSStorages plans and infrastructure.