UniFi Network 10.5: What's New and Why It Matters

UniFi Network 10.5 focuses less on visible interface changes and more on operational reliability. The additions that matter most help prevent configuration lockouts, cut down on unnecessary alerts, and give you better information when a network problem needs diagnosing. Whether any of it matters to you depends on how big and how complex your network is — so here is what changed, who each feature is for, and whether it is worth acting on.

Safe Ops: Configuration Changes That Roll Back If They Break Something

Test and Confirm keeps a configuration change from becoming permanent until your UniFi access points and switches confirm they can still reach the controller. If the change drops connectivity, the previous settings restore on their own after a short auto-revert window instead of leaving you locked out.

We have seen what happens without that safeguard. One client started making his own VLAN changes and ended up breaking the entire network — at 9:00 PM. Because we had a current cloud backup of his configuration, we were able to restore the system remotely that same night, with no emergency site visit. Test and Confirm is built for exactly that scenario: a VLAN or firewall change that would have cut the network off reverts automatically, rather than turning into an after-hours recovery call.

Link Debounce: Less Alert Noise From Brief Link Changes

Link Debounce filters very brief port-state changes before they generate alerts or ripple into topology changes. A momentary blip — the kind a marginal cable or a quick power event can cause — no longer floods the log the instant it happens.

On larger installs, that noise tends to bury the alerts that actually matter. Worth being clear, though: Link Debounce quiets the symptom, it does not fix the cause. A recurring flap from a loose cable or a failing SFP module is a genuine fault, and repeated port flaps still need to be traced back to the cabling, power, or hardware behind them.

Auto STP Edge: Faster Recovery on Endpoint Ports

Auto STP Edge detects ports that connect directly to endpoint devices and lets them skip the slower Spanning Tree Protocol (STP — the mechanism switches use to prevent network loops) listening and learning states, which cuts convergence time and keeps larger topologies more stable. On a single-switch home network this won't be noticeable — but on multi-switch commercial installs, faster convergence means less disruption when a link goes down or a device reconnects.

It does not replace correct switch topology or loop prevention. STP Edge assumes the port leads to an endpoint, not another switch, so if an unmanaged switch or a loop gets plugged into an edge port, the design still has to guard against it — the feature speeds up the common case, it doesn't remove the need to plan the network.

Firewall Hit Statistics: See Which Rules Traffic Actually Hits

Firewall Hit Statistics adds rule-level hit counts directly in the Policy Table, so you can see which firewall rules traffic is actually matching instead of guessing.

That matters most when you have inherited a rule set that has grown and been edited for years. A rule with no recorded hits becomes a candidate for investigation — it might be obsolete, or it might guard an occasional or emergency path that simply hasn't fired during the measured window, so it gets documented or consolidated rather than blindly deleted. Rules that fire constantly, meanwhile, confirm what is actually doing the work day to day.

This is where hit counts earn their place in a network audit. When we review a firewall for a client, we simplify the rule set, tidy up DNS so lookups resolve faster, and separate the network into distinct IoT, security, and main-device segments — and rule-level hit data makes each of those calls with evidence instead of assumption.

Resilient SD-WAN Underlays: Better Multi-Site Reliability

Resilient SD-WAN Underlays are for organizations that connect multiple buildings or locations. SD-WAN — software-defined wide-area networking — manages the links between physical sites as one logical network, and the underlay is the physical connectivity underneath it.

In 10.5, UniFi can build that underlay across direct fibre, MPLS, wireless bridges, or UniFi Building Bridge connections, with improved PPPoE performance layered in, so a failure on one path is less likely to isolate a site. It has little relevance to a normal single-location home network.

Identity-Centric Observability: Troubleshoot From the User's Point of View

Identity-Centric Observability organizes network history around the person using the network rather than around each device. Wi-Fi quality, roaming events, application usage, traffic flows, and connectivity history all appear on one timeline for that user.

So instead of chasing three separate logs for a laptop, a phone, and a tablet to figure out why one person's video calls keep dropping, you follow that person's experience across every device and access point they have touched — a much faster starting point for the intermittent complaints that are otherwise the hardest to pin down.

Expanded Fabric Capabilities: SAML Login for Admins and MSPs

UniFi Network 10.5 adds SAML authentication for Fabric Admins and account-level SAML login for MSP-managed environments. SAML is a standard that lets administrators sign in through an existing identity provider, such as Microsoft Entra ID or Google Workspace, instead of a separate UniFi login.

This is primarily a business and MSP feature with little direct relevance to most home networks — most homeowners can skip it. If you are weighing how far up the networking ladder your business needs to go, we have written a plain-language comparison of how UniFi compares to enterprise network systems.

Should You Update to UniFi Network 10.5?

For most supported UniFi setups, 10.5 is a worthwhile update — the resiliency and visibility features are exactly the kind you want in place before you need them. The sensible caveat applies to any release: setups with custom VLANs, layered firewall policies, multiple switches, or site-to-site links should be backed up and reviewed before the update, since those are the configurations where an unexpected interaction is most likely and hardest to unwind. If you are on straightforward hardware with a simple configuration, applying it through the controller during a quiet hour is usually all it takes.