Inherited a Network? Find the Way Back Before Changing It

Network engineer reviewing switches in a server rack during an infrastructure assessment
In this Field Note
  1. Ask for a transaction, not a verdict
  2. Draw two paths
  3. Prove access before you promise rollback
  4. Give the first change a small contract
  5. Leave with a useful short list

There is a firewall rule named “temporary,” a switch that has been running longer than its warranty, and an unlabeled cable disappearing into the ceiling. You have inherited the network. Naturally, you would like to improve it.

First, make sure you can survive improving it.

The opening hour is for discovering what the organization depends on and what would let you recover a mistake. A successful first visit can end without a single configuration change. It should not end without a better account of how the place works.

This is a method for assessing a functioning or intermittently troubled network with the owner’s permission. During an active outage or suspected compromise, follow the incident lead’s priorities, preserve evidence, and coordinate recovery. A leisurely inventory is not the first duty when the business has already stopped.

Ask for a transaction, not a verdict

“The network is slow” describes a disappointment. “The shipping scanner takes fifteen seconds to accept a label after lunch” gives you a device, an action, a time, and a way to recognize improvement.

Ask someone who does the work to show you. Record what they did, what they expected, what happened, and when it happened. Find out whether one person, one device family, one building, or everybody is affected. Ask what still works during the failure; that answer often narrows the search faster than the complaint.

Also ask about changes outside the IT calendar. More cameras, a new carrier, rearranged shelving, different working hours, or a heater sharing a UPS can matter. The heater rarely attends the change review.

Choose one important transaction to follow first. You can inventory the rest later. A comprehensive list of boxes is a poor substitute for understanding why the shipping department cannot ship.

Draw two paths

The first is the path the service uses: client, access point or switch, gateway, firewall, transport, and application. Add the supporting services it needs, such as authentication, DHCP, and DNS. Include the return path; packets are entitled to a round trip.

The second is the path you use to administer those devices. Where does your VPN terminate? Which switch carries its traffic? Does the management address you intend to change also support your only connection?

Mark every connection as confirmed or assumed. A switch’s neighbor information, interface description, observed traffic, and the cable label provide different kinds of evidence. Compare them. If you have only an old drawing, write “unverified” on it before its neat lines acquire undeserved authority.

Use approved read-only observations first. Active scanning, throughput tests, debug logging, and packet capture need an appropriate scope; they can create load, expose sensitive data, or disturb fragile equipment. Industrial networks require particular care. NIST’s OT security guide explains why their safety and availability constraints need to shape the investigation.

Prove access before you promise rollback

“We have a backup” leaves several questions unanswered:

  • Does an authorized person have a working administrator account?
  • Can that person obtain the credentials if the normal VPN or identity service is unavailable?
  • Is there an independent console or management route, or must someone travel?
  • Does the backup belong to this device, and what is its timestamp?
  • What software, licenses, certificates, keys, and interface mapping does restoration require?

Open the inventory and verify those answers. Keep secrets in the approved vault; record the vault reference in the notes. Download a configuration through the supported method and store it securely. Check that it contains the expected system identity and relevant settings.

Know what the file omits. For example, MikroTik’s configuration-management documentation distinguishes a readable configuration export from a binary backup and lists items that text export does not carry, including system user passwords and installed certificates. The right recovery set may contain several protected files. A text file with the right name is not proof that you can rebuild the router.

A restore rehearsal on an isolated spare provides much better evidence than inspecting a filename. Do not connect that spare to production with duplicated addresses while you are finding out whether the restore worked.

Give the first change a small contract

Once you understand the problem, write a short change card. Here is a hypothetical example, not a prescription for an unknown firewall:

Field Example
Purpose Permit the approved scanner network to reach the new label service.
Evidence Client attempts coincide with denied sessions to the confirmed destination and service.
Scope One narrowly defined policy change; no unrelated cleanup.
Success An operator completes the agreed test transaction; existing printing and management access still work.
Stop condition The test fails, unexpected traffic appears, or management access becomes unreliable.
Recovery Restore the previous policy through a verified access path; name the person authorized to do it.

Set the decision time in advance. If a thirty-minute window requires ten minutes for recovery and verification, minute twenty is a decision point, not permission to try one more interesting idea. Use the device’s supported rollback mechanism where appropriate, but learn its limits on a lab system first. An automatic safeguard does not replace independent access.

Leave with a useful short list

Keep three columns: confirmed findings, unresolved questions, and next actions. Separate business impact from cosmetic untidiness. An undocumented rule that supports an essential service deserves investigation; an exposed management interface or unusable recovery path may deserve much earlier attention.

Write each action so somebody can approve and finish it. “Test restoration of the branch firewall on the spare and record the missing dependencies” is a job. “Improve resilience” is a meeting topic.

If you are doing this yourself, start with one service path, one critical device, and one recovery test. Those three pieces often reveal where the rest of the work belongs. The recovery manual can grow from what you actually verify.

An Oaktown Tech assessment is useful when the dependencies cross several systems, access ownership is uncertain, or the consequences of a wrong move are expensive. Bring the symptom and whatever records you have. You do not need to tidy the network before asking for help; finding out what can safely change is part of the work.

Share this article

LinkedInFacebookXEmail

Field discussion

What would you test next?

Ask a question, challenge an assumption, or share a lesson from the field. Keep customer names, live addresses, credentials, and sensitive network details out of public comments. Charles reviews every comment before it appears. See the privacy notice for how comment information is handled.

Join the discussion

Your comment will appear after Charles reviews it. Required fields are marked .

A pseudonym is fine.

Working through a similar problem?

Describe the network, the difficulty, and the outcome you need. Charles reviews the brief and helps establish the right scope.