NETGEAR is aware of a growing number of phone and online scams. To learn how to stay safe click here.
Forum Discussion
Feature Request - Dynamic bandwidth management for QoS
I have a 4G connection as my main internet connection. Even with a professionally installed directional antenna the available bandwidth heavely fluctuates depending on the time of the day (most likely due to varying load on the base station) or weather conditions. This renders QoS functionality almost useless. To make it work I have to set the maximum bandwidth to lowest bandwidth ever available or I have to set the antibufferbloat percentage really low.
It would be great if the DumaOS would be able to measure the true available bandwidth and dynamically adjust the antibufferbloat limit. The measurement could/should be a small "speedtest". It would also be great if the measurement history could be viewed. This information could be then used for making requests for the operator to add base station capacity. DumaOS could of course one day utilise the same information to make AI based changes with fewer measurements.
As a side note, I'm a long time Netduma R1 user who recently upgraded to XR500. When we were forced to move from ADSL to 4G, the R1 became for some reason unusable (locking frequently, high cpu usage) both with the original fw and the DumaOS beta. Upgrading to XR500 solved all problems and the XR500 has been now running without problems for weeks in a row.
BR, Henri
3 Replies
- A speedtest by its very nature will download as fast as it can, potentially causing you congestion so if this were to happen a lot in order to keep the bandwidth up to date this could cause more issues than it solves. I do understand the request though, I believe it's been discussed perhaps a daily test to keep it accurate but doesn't sound like that'd be ideal for you. What is the low-high bandwidth you generally get?
The bandwidth fluctuates between 5-45Mbps downlink. I think the usage patterns on the base station might be quite stable so a schedule based setting might even work.
Fully understand the potential problem related to "speedtesting" frequently. But I also think such a test could be very short/small; the test performed by Speedtest is overly long. It would be ideal if this measurement could somehow be derived from the natural traffic flow. The problem is that the router cannot know whether a true maximum is achieved with natural traffic.
Could the first thing (MVP) be a configurable periodic speedtest with a quality partner having geodistributed service? And to have a history trend of these results? With these you could at least understand if you have such a problem and it's magnitude. I do assume that mobile connections are becoming very popular so this will be a common issue. At least here in Finland they are dismantling everything else physical than fibre and 4G is offered where fibre is not available (a lot of places).
- Okay that's quite a range, couldn't really put an average in bandwidth settings with that. I think at some point we do want to have more statistics regarding what is happening with your connection so this could form part of that. Good ideas and I will pass them on to the team.
