Backups in Production
Strategies, Tools, and Verified Restores.
Planned
A backup only counts once a restore from it has succeeded. This book will cover what that takes on production systems: deciding what to copy and how often, laying copies out along the 3-2-1 rule, capturing databases and running services consistently, keeping an offsite copy that a compromised host cannot delete, and running restore drills that prove the whole chain.
Status: planned title. Writing has not started. The repository and the cover exist; the scope below is the plan.
The first edition will ship DRM-free as PDF and EPUB, with free updates.
From the first copy to a verified restore.
No chapters are written yet. This is the plan: the areas the book will cover, in the order a reader would meet them. It targets self-managed servers and open source backup tools.
- Strategy and policy Recovery point and recovery time targets, what to back up and what to rebuild from configuration, and retention schedules that match how data ages.
- Tools and their trade-offs File-level tools such as restic and Borg, filesystem snapshots, and image-level backups, with the properties that decide between them.
- Databases and application state Consistent captures of PostgreSQL, MySQL, and other stateful services, dumps compared with snapshots, and point-in-time recovery.
- Offsite and immutable copies Object storage targets, encryption at rest, and append-only modes that keep a copy safe when the source host is compromised.
- Verified restores Restore testing as a scheduled routine, integrity checks on repositories, and drills that measure real recovery time against the target.
- Operations Scheduling, monitoring for missed and failed runs, pruning and repository maintenance, and capacity planning for growing datasets.
For engineers who own the recovery plan.
This book is for people who run servers and answer for the data on them. It assumes comfort with the command line and explains the reasoning behind each layout rather than listing recipes.
- Sysadmins and platform engineers who inherit a backup setup and need to judge whether it would survive a disk failure, a deleted database, or ransomware.
- Teams replacing ad hoc scripts with a managed system that has retention, monitoring, and an offsite copy built in.
- Operators who have never run a full restore and want a drill plan before an incident forces one.
Planned.
Join the shared Sysinit Press book list. You'll get one message when writing on this book starts, plus release news when it ships. The first edition will include DRM-free PDF and EPUB files with free updates.
The signup uses double opt-in. Every message includes an unsubscribe link.