Teams usually back up the production database and forget everything needed to run it again: infrastructure definitions, pipeline settings, secrets and keys, release artifacts, SaaS configuration and documentation. Walk through the list below and mark what could not be rebuilt from memory.
What official guidance includes
CISA's Stop Ransomware Guide goes beyond data. It advises keeping infrastructure as code template files in offline backups so cloud resources can be redeployed quickly, storing source code or executables with offline backups, and maintaining golden images of critical systems. AWS's recovery guidance similarly says that, in addition to data, you must redeploy infrastructure, configuration and application code, and that without infrastructure as code recovery may be slow enough to exceed your RTO.
The checklist
| Area | Examples | Question |
|---|---|---|
| Data | Databases, object storage, queues | Can you restore to a point in time? |
| Code | Repositories, wiki repos | Is there a copy outside the host? |
| Infrastructure | Templates, state, network settings | Could you rebuild the environment? |
| Pipelines | Workflow files, runner images, variables | Could you rebuild CI from scratch? |
| Artifacts | Release binaries, images, packages | Do you hold every shipped version? |
| Secrets and keys | Secret store, signing keys, backup keys | Is there a protected recovery path? |
| Identity | Directory, SSO settings, groups | Can you restore access policy? |
| SaaS settings | Ticketing, monitoring rules, alerts | Are settings exportable? |
| Knowledge | Runbooks, contacts, diagrams | Can you read them when the wiki is down? |
How to find gaps
- Imagine the production account and your Git host are both gone tonight.
- List what you would need on the first day: code, build steps, keys, domains, certificates, customer data.
- For each item, name where the backup is, who can restore it and when it was last tested.
- Mark items with no answer. Those are your gaps.
- Rank gaps by how long they would block recovery, and fix the top three first.
Often missed items
- Runbooks stored only in the system that is down. Keep a printed or offline copy of the first hour steps and contact list.
- DNS records and certificates. Export zone contents and know who controls the registrar account.
- Signing and encryption keys. Losing them can make a backup unreadable. Keep protected copies separately from the backups they unlock.
- CI variables. Often stored only in the CI service. Keep a secure record of names and owners, even when values live in a vault.
- Third party dependencies. Packages and base images you do not host. See mirrors and caches.
Do not back up secrets carelessly
More backups mean more copies of sensitive data. Encrypt them, limit access and set retention.
Safeguard's asset discovery and SBOM generation help list what you run and depend on, and its compliance automation helps keep evidence of backup controls. Try the free scan.
Try it yourself
Click through it at your own pace, or open it full screen.
See how many components a typical project pulls in. Each one is something your build needs, and something to include in the scope.
Quick check
What is the single most forgotten item?
It varies, but the recovery runbook and the keys that unlock backups are common. Both are useless if they sit only in the failed system.
Do SaaS tools back themselves up for me?
Providers protect their platform, but your settings and content may not be recoverable after your own deletion. Check each provider's documentation.
Sources
Last reviewed: 19 October 2026.
