Write the responsibility matrix
A dedicated server may include hardware replacement but exclude operating-system recovery, data restore, firewall policy, or application diagnosis. List each layer and the response expected from provider and customer.
Confirm replacement stock, response targets, remote console access, drive handling, and what happens when diagnostics are inconclusive.
- Chassis and components
- Firmware and operating system
- Network policy
- Backup and restore
Plan beyond one chassis
Local redundancy does not cover chassis loss, facility isolation, account suspension, or operator error. Keep recoverable artifacts and data outside the server's failure and billing boundary.
Capacity decisions should use measured CPU, memory, storage latency, network, and growth rather than a generic server tier.
Recover the service onto different hardware from documented configuration and external backups.