Single-Uplink Failure Modes in Small Office Networks

My name is Tony Underwood. I own Cambium Data in Omaha, and I still see professional offices that run the entire day through one internet circuit with no tested path for failure. Phones, cloud applications, practice-management systems, email, scanners, VPNs, and security devices may all depend on the same handoff. When that path fails, the office can lose more than web access.

A single circuit is not automatically a bad design. The risk is treating it as redundant when it is not, or assuming that a backup path will work without testing the complete failure and recovery process.

The first step is to map the dependency chain. Identify the provider handoff, edge router, firewall, switch stack, wireless controller, DNS path, VPN termination, voice service, and any cloud applications that staff need to do their work. A second circuit does not help if both circuits terminate in the same failed device or use the same physical route.

Questions I use when reviewing a small-office uplink include:

• Is there a second circuit from a genuinely separate path?
• Could a cellular connection carry the most important traffic during an outage?
• Does the edge device fail over automatically, or does someone have to move a cable or change a setting?
• Will DNS, VPN, voice, and authentication continue to work after the transition?
• Does the backup have enough capacity for critical applications, or only for basic email and browsing?
• How does the team know that the backup path is carrying production traffic?
• What is the intended recovery order when the primary circuit returns?
• Who receives the alert, and who is responsible for confirming that service is normal again?

Failover also needs a controlled test. Disconnecting a cable and seeing a link return is not the same as confirming that users can reach the applications they depend on. A useful test checks name resolution, VPN sessions, voice, cloud applications, scanners, monitoring, and notification. It should also record how long the transition takes and what has to be restored manually.

Recovery deserves the same attention. If the primary circuit returns, automatic failback may interrupt active sessions or send traffic back to a path that is still unstable. Some networks should fail back immediately. Others should wait for a defined health check and an operator decision. The right choice depends on the applications and the failure mode.

The practical lesson is simple: document the single point of failure, decide which services matter first, and test the backup path under conditions that resemble a real outage. The goal is not an impressive diagram. The goal is a predictable response when the upstream path disappears.

I am Tony Underwood, CCIE #7112. I started in the Navy and moved into networking and IT from there. I have owned Cambium Data since 2009, working with legal firms, dental practices, and other professional offices in the Omaha metro, usually with 15 to 120 employees. These observations come from network reliability work, not a promise that every office needs the same design.

More about me: https://site.tonyunderwood.us/
LinkedIn: tonyunderwoodomaha

Author contact:
Cambium Data
6542 S 118th St, Omaha, NE 68137
[phone]
Tony Underwood
[e-mail]