Everything we know about
keeping databases recoverable
Field notes, playbooks and guides from running backup and monitoring infrastructure for MySQL, PostgreSQL, SQL Server and MongoDB.
What Real-Time Database Monitoring Should Actually Measure
CPU and RAM are table stakes. The signals that predict an outage are connection saturation, replication lag and p95 query latency.
Read articleWhy an Agent Beats Giving a SaaS Your Database Credentials
Outbound-only agents, zero-knowledge encryption and read-only metadata access — the connection model behind SaveMyDB fetching.
Auto Discovery: Finding the Databases Nobody Documented
The database that takes you down is usually the one missing from the spreadsheet. Continuous discovery keeps the inventory honest.
The 3-2-1 Backup Rule for Production Databases
Three copies, two media types, one off-site. How the classic rule maps onto MySQL, PostgreSQL, SQL Server and MongoDB workloads in 2026.
PostgreSQL Backup Best Practices That Actually Scale
Logical dumps, physical base backups and WAL archiving — when to use each, and how to keep restore times predictable as your cluster grows.
RPO and RTO Explained Without the Jargon
How much data can you afford to lose, and how long can you afford to be down? Two numbers that decide your entire backup schedule.
Five Monitoring Signals That Predict Database Outages
Connection saturation, replication lag, slow query drift, disk headroom and lock waits — the early warnings worth alerting on.
A Restore Drill Playbook Your Team Can Run in 30 Minutes
An untested backup is a rumour. A short, repeatable drill that proves your recovery path works — and documents it for the next person.
Self-Hosted Vault vs Vendor Storage: Who Holds Your Data?
Backup providers that store your data on their infrastructure trade control for convenience. Here is what that trade actually costs.
Agent installation
One-line install for Linux and a signed installer for Windows. Outbound-only, no inbound ports required.
Vault configuration
Connect your own object storage, set retention windows and rotate encryption keys on your schedule.
Restore runbook
A step-by-step recovery path you can hand to anyone on call, from snapshot selection to verification.
