Connect once. Prove it every week.
DrillProof is a control plane, not a destination. It orchestrates backups and restores inside your own cloud account through a role you grant and can revoke — and it keeps proving the restores work long after everyone has stopped thinking about them.
- 01
Connect your cloud, keep your data
You create an IAM role in your own account with a trust policy naming DrillProof and an external ID unique to you. That role carries only the permissions needed to snapshot, restore, and clean up — no console access, no long-lived keys, nothing that can read application data at rest.
DrillProof assumes that role to issue API calls. Because the work happens inside your account, snapshots and restores never traverse our infrastructure. Delete the role and the integration stops immediately; there is nothing else to revoke.
- 02
Back up immutably, tier automatically
Cluster state and manifests are captured through Velero; persistent volumes are snapshotted on EBS; managed databases are captured on RDS. These are grouped into a single restore point, so a recovery brings back a coherent application rather than three unrelated artifacts.
Restore points are written to S3 under Object Lock, which enforces WORM retention at the storage layer — a compromised credential cannot shorten retention or delete history. As restore points age, lifecycle policy moves them down the storage classes so long retention stays affordable.
- 03
Drill on a schedule, get proof
On the cadence you choose, DrillProof restores the latest restore point into an isolated environment in your account, asserts that it actually works, then destroys the environment. Production is untouched, and the cost is the runtime of the drill.
What remains is evidence: what was restored, which checks passed, the observed RTO, and a timestamp. Failures surface as alerts with the reason attached, which is the entire point — a drill that fails in a sandbox on a Tuesday is a good outcome.
Where everything actually lives
The boundary is the product decision. DrillProof is built so that losing us costs you nothing and reaches nothing.
DrillProof control plane
Schedules, policies, and drill results. Metadata only.
- backup + drill schedules
- retention & tiering policy
- drill outcomes, timings
- evidence records
No snapshots, volumes, or database contents are stored here.
AssumeRole
+ external ID
drill result
+ metadata
Your AWS account
Every byte and every restore stays here, in your region.
Workloads
EKS · EBS · RDS
Backup vault
S3 Object Lock
Drill sandbox
created, verified, destroyed
The drill sandbox is dashed because it is temporary: DrillProof creates it to prove the restore, then destroys it.
Drill lifecycle
What happens during a drill
Four stages, every time, with a record at the end. This is the loop that turns a backup into a proven recovery path.
- 01
Restore
The latest restore point is recovered into an isolated, throwaway environment in your account. Production is never touched.
- 02
Verify
Volumes mount, the database accepts a connection, and workloads reach ready. Checks are asserted, not assumed.
- 03
Tear down
The isolated environment is destroyed, so a drill costs the runtime of the drill and nothing after it.
- 04
Evidence
A dated record of what was restored, what was checked, and how long it took — retained for auditors.
A drill leaves a log, not a green tick
Every stage is timestamped and every assertion is named. When an auditor asks how you know recovery works, this is the answer — and when a drill fails, the same log tells you exactly which assumption broke.
$ drillproof drill run --plan prod-cluster-weekly 10:02:14 restore creating isolated namespace drill-7f3a in your account 10:02:51 restore orders-pv 12.4 GB restored from snap-0e1c9 10:03:36 restore payments-db 41.7 GB restored to drill-7f3a-rds 10:05:02 verify volume mounted, filesystem clean 10:05:19 verify database accepting connections, 1 query returned 10:05:44 verify 8/8 workloads reached ready 10:05:58 teardown drill-7f3a destroyed, 0 resources remaining VERIFIED RECOVERABLE RTO observed 3m 44s evidence ev_2f81c4 · retained 12 months
What we cover today
Stated plainly, because discovering a gap after signing is worse than reading it here.
- Amazon EKS
- Cluster state and workload manifests
- Amazon EBS
- Persistent volumes behind your pods
- Amazon RDS
- Managed relational databases
Not yet, and we will not pretend otherwise
Amazon EFS, DynamoDB, S3, SaaS sources are on the roadmap and not shipped. DrillProof supports AWS at launch — we are not multi-cloud, and anyone who tells you their day-one product is has not run a cross-cloud restore.
See it against your own backups
The free audit runs this process once, on your latest restore points, and gives you the report.