We used a $1M tool to test the HOT $279 Ubiquiti UCG-Fiber Gateway

Ubiquiti Cloud Gateway Fiber Review

Patrick from ServeTheHome reviews the Ubiquiti Cloud Gateway Fiber (UCG-Fiber), a compact router and firewall. He tests it with Keysight CyPerf, a costly traffic-generation system capable of more than 1.6 Tbps, to examine its hardware and performance under realistic application traffic—not just simpler tests such as iperf3.

Hardware and features

The UCG-Fiber has a front LCD, two 10GbE SFP+ ports, a 10GbE RJ45 port, and four 2.5GbE ports. The 10GbE ports are configured as WAN ports by default. One 2.5GbE port provides PoE output for an access point; the reviewer would like to see PoE on more ports.

A Fluke PoE tester negotiated only about 6.5 W (Class 2), rather than PoE+, across three units. The presenter notes that the negotiation may be unusual and that the device has provided more power in other circumstances.

Storage options include a model without storage, a 1TB version priced at $399, and a 2TB version priced at $529. An optional $19 kit supports a customer-supplied M.2 drive.

Inside, the gateway has a quad-core ARM Cortex-A73 processor running at about 2.2 GHz and 3GB of memory. Its cooling fan is glued in place and difficult to access, making cleaning or replacement inconvenient.

The unit draws around 8–9 W at idle. The fan was generally quiet during testing, measuring roughly 35–37 dBA in the studio.

UniFi software integration is a major strength. The gateway can act as a controller for access points and run services such as Protect/NVR, using its SSD to store camera recordings.

Performance and testing

A simulated 500-person Zoom meeting—with five presenters and about 490 attendees—ran without reported audio or video problems.

In a mixed workload of applications and up to about 1,500 users and 60,000 concurrent connections, the gateway handled a full-duplex 10GbE link: roughly 10 Gbps in each direction, or about 20 Gbps total at the measured link level.

At roughly 2,048 users, traffic abruptly stopped passing, and the device required a physical power cycle to recover. In a separate run using an 11-application mix, it also stopped passing traffic at about 1,536 users. The presenter suggests system memory, buffers, or other resource limits may be involved. These tests used heavy loads beyond the device’s expected deployment and did not enable the full IDS stack, Protect, or other additional features.

Attack testing showed limited blocking in the default configuration. One run blocked about five of a few hundred attack flows; another test involved more than 5,000 attacks, of which around 70 were blocked. The presenter says a higher-tier threat-protection license may provide additional capability and invites interest in further testing.

The reviewer argues that realistic workloads can reveal behavior that iperf3-style tests miss. As user count and application variety increase, throughput does not necessarily remain at the 10GbE ceiling; the gateway’s relatively low-power processor is a limiting factor.

Overall assessment

The presenter considers the UCG-Fiber a strong, compact option for homes and small-to-medium businesses, especially for users invested in the UniFi ecosystem. The high-load failures are presented as useful information about the device’s limits, rather than evidence that it is unsuitable for its intended use.

The main concerns are the difficult-to-service fan, the unexpected PoE negotiation result, and the possibility that extreme workloads can leave the gateway requiring a hard reboot.

Coverage and sources

The review covers the device’s hardware and teardown, realistic-traffic performance testing with CyPerf, and its strengths and limitations—including PoE behavior and UniFi software features.

Main speaker/source: Patrick of ServeTheHome (STH). Testing and measurements discussed include Keysight CyPerf traffic generation and a Fluke PoE tester.