In this Field Note
Pick one important application. Now explain how you would restore it if the administrator’s laptop, the identity service, and the shared drive were unavailable. The password file on that shared drive is unavailable too.
This is where a reassuring backup dashboard can become less reassuring. The files may exist, but the database needs storage, storage needs credentials, and the credentials depend on the identity system you have not rebuilt yet.
You can untangle those dependencies during a planned exercise or during a very long incident. The planned exercise has better working conditions.
Start with one useful business function
“Recover everything” is an ambition. It does not tell the engineer what to bring back first.
Ask the people responsible for operations what minimum function lets them resume useful work. For one organization it might be accepting and fulfilling orders. For another it could be laboratory acquisition with access to reference data. Name the function, its owner, acceptable downtime, and the amount of recent work the business can tolerate losing.
Those last two decisions are commonly called recovery time and recovery point objectives. Give them operational meaning. “By the next shift, with no more than the last completed batch to re-enter” is a requirement people can discuss. A pair of unexplained acronyms is mostly decoration.
Then list the dependencies: identity, DNS, time, storage, application servers, databases, licenses, endpoints, network paths, and any outside supplier. NIST’s cybersecurity recovery guide emphasizes prioritizing resources and developing realistic recovery plans and tests. Your first version can be a short sequence for one function; it needs to be usable before it needs to be comprehensive.
Make the recovery instructions survive the incident
For each step, write down who performs it, what access they need, where the instructions live, and what proves it worked. Look for circular dependencies: the password vault needs the application server, which needs the credentials stored in the vault.
Use a protected independent way to obtain essential recovery material. That may involve approved emergency access, offline records, or another mechanism suited to the organization. It should be controlled, reviewed, and tested—not a loose copy of every password in somebody’s desk.
Include a known-clean administrative device and a communication method that does not depend entirely on the affected environment. Verify how the authorized team gains access without casually weakening the controls you are trying to recover.
Ask a qualified colleague who did not write the instructions to follow them. Watch where explanation is required. Those gaps are part of the work, even if the original author can solve each one from memory.
Protect the recovery copy from ordinary access
A compromised workstation account should not be able to delete production data and every backup. Separate backup administration from routine work, control privileged access, and maintain suitably protected offline or immutable recovery copies.
Understand exactly what “immutable” means on the chosen platform: which retention applies, who can change it, which administrator can delete the surrounding resource, and what happens when storage fills. A label on a product page does not describe your configuration.
The CISA #StopRansomware Guide calls for offline, encrypted backups and regular recovery testing. It also recommends retaining the software and configuration material needed to rebuild systems. A data copy without the application, license, or supported deployment method may leave considerable work unresolved.
Preserve enough history to assess a suitable recovery point after a compromise. The newest backup may contain malicious changes or damaged data. Selecting a point belongs in the incident process, with the appropriate security and application specialists.
Run a rehearsal with a finish line
Begin with a tabletop if you have never walked the dependencies. Do not disconnect production systems to make the discussion feel realistic. Then schedule a technical restore into an isolated, approved environment with controlled connections.
This is an example exercise for one business application:
- Confirm authorized access from the recovery administrator’s clean device without relying on the normal laptop.
- Bring up the required network, name resolution, identity, and storage dependencies in the documented order.
- Restore the selected application and data, including configuration, permissions, and necessary licenses.
- Have an application owner perform a representative business task and verify expected records.
- Record elapsed time, recovered-data age, failures, workarounds, and tasks requiring undocumented knowledge.
- Assign owners and dates to the gaps, then repeat the failed portions.
The finish line is a usable business function in the agreed environment. “The restore job completed” is an earlier milestone.
For an illustrative order system, that might mean an authorized test user can find an existing order, create a controlled test transaction, and produce the expected output without sending unintended messages or payments. The application owner defines the safe test. A laboratory’s equivalent would be different.
NIST’s data-integrity recovery work emphasizes confidence that recovered information is accurate, complete, and free of malware. That is why user validation and security assessment belong alongside the restore log.
Reduce the work an incident can create
Recovery preparation does not excuse an exposed, unsupported remote-access service. Keep edge services maintained, remove unnecessary exposure, strengthen administrative authentication, and separate routine accounts from privileged work. The NCSC’s ransomware guidance describes these defenses alongside backup and recovery.
On the network, define management boundaries, constrain vendor access, and preserve useful logs. Give alerts an owner and decide who can authorize isolation. A subscription can deliver a notification; it cannot decide which production process may safely stop.
Utilities and laboratories need that distinction in writing. Disconnecting a system can affect observation, control, or safe operation. Coordinate containment procedures with the people responsible for the physical process and the relevant vendors before an emergency compresses the decision.
If the incident is already happening
Activate the organization’s incident plan and involve its designated response specialists. Use known-clean communications, contain affected systems through the safest available method, and preserve relevant evidence. Indiscriminate wiping or rebooting can destroy information the investigation needs.
Before reconnecting restored systems, coordinate assessment of the entry path, compromised identities, persistence, and the recovery environment. Restoring yesterday’s files does not establish that yesterday’s access problem has gone away. Legal notification, insurance, forensic investigation, and ransom decisions belong with the responsible specialists.
Oaktown Tech can help with containment boundaries, protected management, configuration recovery, and rebuilding maintainable infrastructure, working alongside endpoint, backup, and incident-response specialists. Emergency work requires availability and explicit acceptance; this is not an implied 24/7 response or forensic service.
For preparation, start with one application and a real restore. A list of discovered gaps may feel less comforting than a green dashboard. It is considerably more useful.

Join the discussion