Skip to main content

What are you looking for?

Explore our services and discover how we can help you achieve your goals

Client Workspace

Proving Restores Work: A Backup Restore Test Runbook and Evidence Log

You prove a backup works by restoring it into a separate environment, checking that the application runs on the restored data and timing the whole recovery against its target. Netbase's seven-step runbook below does that on a schedule and writes every result into an evidence log anyone can check.

Book a solution review See the related solution

By Netbase Commerce and Platform Engineering Desk · Reviewed by David (CEO) · Updated · 10 min read

star

This guide is for tech leads, operations leads and CTOs who own a live business application and its backups. It sits under Netbase's cloud platform engineering work, where backup policy, restore tests and a recovery plan are part of the platform, and the automation belongs in DevOps and platform engineering pipelines.

In this guide

Why is a successful backup job not proof that you can restore?

A green backup job only proves that a copy was written. It says nothing about whether the copy is complete, whether it can be read back, or whether the application starts on it. Microsoft's reliability guidance warns that backups can be misconfigured to capture only part of the data and asks teams to validate integrity and completeness, not just whether a restore operation succeeds.

AWS lists the same traps as anti-patterns in its recovery testing practice: restoring a backup without querying any data, assuming the backup exists, and assuming the restore fits the recovery time objective. A restore test turns each assumption into a recorded fact with a date.

Two targets frame every test. Google Cloud defines the recovery time objective (RTO) as the maximum acceptable time an application can be offline, and the recovery point objective (RPO) as the maximum acceptable period of data loss. A restore test measures both: how long recovery took, and how old the newest restored record is.

What does a restore test have to prove?

A restore test passes only when four things are true, and each needs its own check:

  • Data

    Question it answers
    Is everything there and readable?
    Check that proves it
    Row counts, checksums or a sample of known records against the source
  • Freshness

    Question it answers
    Is data loss inside the RPO?
    Check that proves it
    Timestamp of the newest restored record against the incident time
  • Application

    Question it answers
    Does the product work on the restored data?
    Check that proves it
    The application starts, a test user logs in and completes one core action
  • Time

    Question it answers
    Is recovery inside the RTO?
    Check that proves it
    Clock from the decision to restore until the application check passes

Microsoft adds that recovery must be measured across the full flow, because one component can restore in time while the whole service still misses its target once dependencies also need restoring. Time the application, not only the database.

Which test depth fits which cadence?

Not every test needs a full rebuild. Pick the depth by what changed and what you need to prove:

Criterion Integrity check Data restore Full application restore
What it proves The backup file or snapshot is readable and complete One database or bucket restores and its data validates The product runs on restored data inside the RTO
What it does not prove That anything restores or starts That the application, secrets and configuration come back Recovery in another region or account
Typical trigger After every backup run On a fixed schedule and after a schema change Before a major release, after an infrastructure change and on a fixed schedule
Who reviews the result Automated alert Engineer on duty Tech lead and the business owner

Microsoft's guidance ties restore cadence to how often the backup configuration, data schema or infrastructure changes: the more often they change, the more often restores need testing. Point-in-time recovery deserves its own test: PostgreSQL's continuous archiving needs an unbroken sequence of archived WAL files back to the base backup, and its documentation asks teams to set up and test WAL archiving before taking the first base backup.

The restore test runbook

  1. Name the scope and the pass criteria

    Write which system, which backup and the four checks above before you start, so nobody changes the bar during the test.

  2. Pick the recovery point

    Alternate the latest backup with an older or random one, since a recovery may need an older copy.

  3. Start the clock at the decision

    Record when the restore was declared, then who found the runbook, the credentials and the access.

  4. Restore into an isolated environment

    Use a separate account or project, never production, with its own network and no outbound mail or payment calls.

  5. Validate data, freshness and the application

    Run the data checks, read the newest record time and complete one core user action.

  6. Stop the clock and compare

    Note the elapsed time against the RTO and the data age against the RPO, and flag any gap.

  7. Log, fix and tear down

    Write the evidence log entry, open a task for every gap with an owner, and delete the restored copy.

A person other than the backup owner should run the runbook at least some of the time. The test then checks the written procedure, not one engineer's memory.

What should the restore evidence log contain?

The evidence log is the record a buyer, auditor or insurer asks for. Keep it outside the backup tool, so it survives if the tool or account is lost. One entry per test:

  • Test id and date

    Restore test, month and system name

  • System and backup

    Application, database or bucket, recovery point id and its timestamp

  • Depth

    Integrity, data restore or full application restore

  • Operator and reviewer

    Role of the person who ran it and the person who signed it off

  • Declared, restored, validated

    Three timestamps: decision, restore finished, last check passed

  • Result per proof

    Data, freshness, application and time: pass or fail with the evidence link

  • Gaps found

    Missing secret, wrong permission, stale runbook step, missing dependency

  • Actions and owners

    Task reference and owner for every gap, with a due date

A failed test is a useful entry, not an embarrassment. The log shows that gaps are found and closed, which is what a restore programme is for.

Can the cloud provider run restore tests for you?

Partly. AWS Backup restore testing runs scheduled restores of supported resources such as Amazon RDS, Aurora, DynamoDB, EBS, EFS and S3, picks either the latest or a random recovery point from a window you set, records how long each restore job took and then deletes the restored resources. You can retain the restored copy for 1 to 168 hours to run your own validation first, and AWS recommends a dedicated account for restore tests.

What the provider cannot do is decide whether your application works. A managed restore test proves that a resource comes back; the application check, the secrets and configuration, and the clock that starts at the decision remain your runbook's job. Use provider automation for the data restore depth and keep the full application restore as a human-led exercise.

When is a full application restore not worth it?

Do not run a full application restore every week for a small product with one database and a managed backup service. A scheduled data restore with automated checks covers most risk, and a full restore before major releases and after infrastructure changes covers the rest.

Do not build cross-region restore drills before single-region restores pass reliably. If the basic restore misses its RTO, a more complex test only adds noise. Fix the basic runbook first, then decide on a disaster recovery tier.

Where AI fits in restore testing

AI helps with the paperwork around a restore, not with deciding that it passed. An assistant can draft runbook steps from infrastructure code, compare restored row counts and schema against the source and summarise a failed restore job's logs into a first gap note. Netbase works with the major commercial and open-source AI tools and models, chosen per project.

Engineers still define the pass criteria, run the restore, sign the evidence log entry and own every fix; no model gets rights to delete backups, change retention or restore into production.

What delivery record exists, and what does not

Netbase names AWS, Google Cloud, DigitalOcean and Cloudflare as the cloud providers it uses. It also works as offshore development partner on a multi-tenant cloud ERP delivered as SaaS to small and mid-sized businesses for a US client, not named.

Delivery runs in a six-step lifecycle with weekly reviews and KPI dashboards, and most Netbase projects are delivered on fixed-price contracts. No published Netbase record describes a restore test programme in detail, and no restore duration, recovery target or result figure is given here.

Alternatives: who runs the restore tests?

Model Fits when Selection criterion
Your own engineers on a calendar A team with time reserved for upkeep Restore tests survive busy feature sprints
Provider automation plus a quarterly human review Supported managed databases and storage Someone still runs the application check and reads the log
A managed service partner No in-house operations capacity The partner shows you its evidence log, not only a backup report

Restore testing usually sits inside a wider care routine; Netbase's managed cloud services list restore tests as part of the scope, and the patch, scan and recover cadence shows how restores fit next to patching and scanning. To score restore practice against the rest of delivery, use the DevOps maturity scorecard. During a move, the final backup becomes a rollback point in the cutover plan.

What are the limits of this guide?

  • It covers restore testing for one application, not choosing a disaster recovery tier or multi-region failover.
  • Provider features and limits change; check the current documentation before you automate.

Plan the next step with a Netbase consultant

Frequently asked questions

Run an integrity check after every backup, a data restore on a fixed schedule and after schema changes, and a full application restore before major releases and after infrastructure changes. Teams whose configuration changes often need to test more often.

No. A restore test proves that one backup brings one application back within its targets. A disaster recovery test also covers failover to another region or account, traffic switching and failback, and it builds on restore tests that already pass.

They ask for proof that restores were tested, not only that backups ran. A dated evidence log with the recovery point, the elapsed time, the checks performed, the gaps found and their owners answers that question.

Next step

If you want your backups proven before an outage asks the question, book a solution review with Netbase.

AI-ready cloud platform engineering for SaaS and commerce workloads AI-ready cloud platform engineering for SaaS and commerce workloads

Netbase provides cloud platform engineering for SaaS and commerce teams that need a secure, repeatable home for their software and its AI features, on AWS, Google Cloud, DigitalOcean or Cloudflare. We design the architecture, write the environments as code, connect managed and AI services and hand over a platform your team can run, built the way we run ours.

Learn More
line
Managed cloud services with AI-assisted monitoring that keep your platform healthy Managed cloud services with AI-assisted monitoring that keep your platform healthy

Netbase provides managed cloud services for companies with a live platform to keep it secure, current and affordable after launch: AI-assisted monitoring and alert triage, patching, backups and regular cost reviews on AWS, Google Cloud, DigitalOcean or Cloudflare. Our team supports you Monday to Saturday, with Sunday off. SLAs are agreed on request; we publish no standard SLA figures.

Learn More
line
AI-assisted DevOps and platform engineering for routine releases AI-assisted DevOps and platform engineering for routine releases

Netbase provides DevOps and platform engineering for SaaS and commerce teams to make releases routine instead of risky. We set up CI/CD pipelines, repeatable environments and observability on AWS, Google Cloud, DigitalOcean or Cloudflare, with AI assistance for tests, reviews and incident triage, so every change ships the same way and problems surface before customers notice.

Learn More
line
SaaS product accelerator: launch an AI-ready SaaS on proven Netbase modules SaaS product accelerator: launch an AI-ready SaaS on proven Netbase modules

A SaaS product accelerator is a set of reusable Netbase modules for accounts, billing, roles and integrations that helps founders and product teams launch subscription software faster, with room for in-product AI from the first release. Reusing these modules can shorten development time, and the approach is proven on Printcart, Netbase's own web-to-print SaaS.

Learn More
line

Work with Netbase remotely

Every Netbase engagement is delivered remotely, so clients in any country and any industry can work with us; Netbase has worked with clients in 80 countries. Contracts and invoices are issued by Netbase JSC, and payment is made by bank transfer to Netbase JSC's company account or PayPal Business. Talk to the CEO, Huy Nguyen (David), on WhatsApp or by email.

Contact Netbase

Discuss a project

Netbase JSC helps organizations design, build, modernize, and operate digital products and AI-enabled business systems.
Project enquiries

project@netbasejsc.com

WhatsApp

+84 937 869 689

Office address

91 Nguyen Chi Thanh, Dong Da, Hanoi, Vietnam

Get in touch

Tell us what you want to build, modernize, or operate.

Tell us what you want to build, modernize, or operate.

Contact Netbase