I've noticed a lot of people on the forums complaining that their speedtest results don't show the full potential of their connection unless they overstate it in the streamboost setup. I.E. someone with a 40/2 connection won't see 40/2 in a speed test unless they put 50/3 or something higher into SB.
If you're putting anything other than your actual* speedtest results into SB you may as well not even bother switching it on. (*actual = direct result obtained without streamboost or anything else affecting the result so you know its the real world maximum speed your connection is capable of)
The main goal of any QoS setup, including SB, is to prevent your internet connection from hitting 100% bandwidth (saturation) in either direction. When SB is on it will deliberately prevent any single device on your network from being able to utilise 100% of the downstream connection; when you have a 40/2 connection and see something like 32/1.4 with SB on, that means SB is working and you're going to get the results you want when a bunch of devices are all trying to be greedy with the internet at the same time.
Specifically with gaming, lag spikes are due to your connection momentarily hitting saturation in either direction. There is very little bandwidth needed for gaming, but if the internet connection ever hits 100% then the packets are gonna get queued up either trying to get in or out of your network and bam you have lag until that queue is cleared.
Your router can completely control every packet that goes out to the internet - they go from your device, to the router, then the internet.
Your router can only at best indirectly influence the packets that come in from the internet - they are often coming in as a response to a request from a device on your network, so the best the router can do is slow/stop the requests so that no more responses come back.
QoS creates a buffer in the downstream so that the internet connection doesn't saturate in between the time the router realises that it needs to slow/stop upstream requests, does the necessary upstream throttling/re-prioritisation and the effects of that are felt in the downstream. 15-30% is the recommended size of this buffer for both direction in other QoS systems I've worked with. SB seems to have a good engine which reserves a lower amount in the upstream, probably because it can analyse the traffic, knows what needs to go out to the internet right now and what can wait, and has the ability to directly control those packets.
If you have a 40/2 connection, dialled in at around 40/2 as per your speedtest results and you're seeing anything up to about 30% loss of speed with SB on then that's totally normal - that's just how QoS works. Even though SB is a proprietary QoS implementation, it will work like every other QoS system at its core; the automatic identification and categorisation of traffic is what makes SB better than other QoS systems, particularly the ones where you have to manually define priorities for traffic on different ports, etc. I found it very hard to consistently trap torrent/p2p traffic on a manual QoS system but SB is spot on at putting almost everything in the right class.
If you set your SB to above the proper values so you see a full speed speedtest while its on, you're actually getting no protection from saturation of your connection, which is what causes lag and all the other annoying things you probably bought this router to avoid. I think there needs to be a sticky about QoS
Author
Topic: SpeedTest Results... (Read 38284 times)