450M - Wireless Speed Tests great... Actual Speeds Poor

Hey Everyone, 

Were a few months into our 450M deployment and were seeing some strange issues.

Customer signals, SNR, and Modulation look great. We run wireless speed tests and get 100+Mbps. When we try a local network speed test we see under 10Mbps. We checked QOS settings and set them to MAX possible. We have verified capacity at the tower, and we have well over 1000Mbps when running any type of test. 

The issue is odd, when running wireless link test with Bridging we see the expected 100Mbps plus speeds, when trying a speedtest with Bridging and MIR we see near 10-15 Mbps, and on an actual speed test (speedtest.net and fast.com) we see only up to 10Mbps. Not sure if this is an AP capacity issue or a bug on QOS, or a configuration issue. Would love some help. See attached screen shots.



Are you sure that none of your other SM's have priority set?

Yes, all are equal.


we used to face this problem in our network. Radios were PERFECT, yet actual speed tests were bad and customer's weren't too happy.

we ended up fixing the problem, but it took a long range plan to do so. what killed us was the multicast/unicast traffic, so we designed a new VRRP and segmented customer traffic into multiple dhcp vlans on the 450m (no more than 25 customers on a vlan), and this SOLVED the issue for us. 


We have this exact issue on a 450m 3ghz ap that replaced a 450 ap with 60 subs on it. The 450m actually performs worse and it doesn’t matter how much load is on it or if the sm is grouped or not. The sm might test 100Mbps with only 50Mbps load on the ap yet the customer can only pass 1 to 10Mbps with the QOS on the sm set to the max low priority QOS. There are only a few high priority channels set for voice but these are low demand. Customers are complaining, beta firmware causes the ap to fail with some obscure error and a Cambium engineer was on site last week and witnessed the issue. We’re at a loss. The ap won’t pass over 80Mbps and this was the max of the standard 450.

We had problems like that when our upstream providers or who they were required to partner with were clearly identifying our customers traffic and even where they lived and we observed different symptoms even with things like if a cell tower was planned on being deployed in an area or not.  If you haven't done so already please make sure how you are testing speeds is not being done in such a way where outside of your network identifying and shaping of your customers traffic could be happening.  Not having any experience with 450m gear myself I can only speculate and offer some experience from things similar to what you are describing wheh we thought the 450i and 450 gear should be passing more traffic than the 430 gear was and in retrospect it was clearly illegal identifying and shaping of our customers traffic outside my network.

From digging more into the interface it seems we’re hitting a maximum on specific sounding areas. So not the entire AP is maxed out but some specific areas. Our next step is to deploy a second AP in the same area and migrate slow customers to it to balance the two out.

1 Like

3 posts were split to a new topic: 450M performance problems with 16.2.2