I think I’m clear on how the “Cut-off latency” and “Suspension Time after Packet Loss” features function, but I haven’t been able to find much of an explanation of how “WAN Smoothing” functions (Off / Normal / High).
Could someone explain what “WAN Smoothing” does and how (if at all) it relates to the other two settings mentioned above?
Thanks!
WAN Smoothing works quite independently, and require you to have at least 2 WANs on either side to be effective. The concept here is to reduce the impact of packet loss and get the lowest possible latency, at the expense of extra bandwidth consumption. If you set this option to Normal, currently it will consume 2 times the required bandwidth, for example, if you have a 1Mbps video stream, SpeedFusion with WAN Smoothing will send out 2Mbps (max.) of data, using some dynamic algorithm to add extra bits for data correction. As we are still enhancing this feature, so the overhead or algorithm used is subject to change in the future.
If you have one peer (A) with one very very fast internet connection, and another peer (B) with multiple slow connections and suppose you are primarily concerned about the smoothness of the throughput from A to B (sufficient stable capacity is generally available from B to A, not so much from A to B) – then will turning on WAN smoothing on peer A’s speedfusion profile configuration cause a desirable effect (eg cause A to send more redundant packets to each of B’s wan’s than it would without turning the setting on)?
Thanks,
Hi,
You should enable WAN Smoothing on site A if site B has a healthy WAN link.
Hope this help.
Is WAN smoothing fairly dynamic in that it will scale up and down (to the level you have set) according to needs? So basically can I set it at MAX and not have to worry about wasting bandwidth if circumstances are not needing a lot of error correction? For example, if conditions are good maybe it operates at NORMAL level even if set on MAX level. Or will MAX always use roughly close to 4X bandwidth?
The other thing I have wondered is if remote connection always waits for error correction bits? Or are error correction bits only looked for when needed?
WAN Smoothing works based on the defined setting. For example, bandwidth will be consumed as 2x of the original data traffic if Normal was chosen.
Hmm… OK. In the description I think it says it will use “AT MOST” 2x 3x – so that led me to believe that it wasn’t exactly 2x etc… seems to imply that it isn’t 1:2 relationship and/or that there is some dynamic element. Can you elaborate on how it works? Also, I noticed that under “HIGH” setting it says that the bandwidth used depends on the number of WAN’s connected – it doesn’t say 1:4 relationship like the ratios it shows for normal/medium setting. Will “high” be roughly 1:4 or is it even higher, say if I have 5 WAN’s connected?
If you don’t have dynamic in the existing settings then maybe you could add this as a future feature so it would scale up and down based on if it conditions need more or less. That would make the feature far more appealing to me as I hate to waste bandwidth if it isn’t helping. (That’s another thing that would be helpful – some kind of graph or data point showing how much smoothing is being used, maybe in the number of errors corrected per second/minute etc? That way a user could at least know if having it on HIGH setting was dumb because not needed?)
Please find the drawing below. This is how the SpeedFusion tunnels look like if both side have 3 WANs.
WAN Smoothing - Off
WAN Smoothing - Normal
WAN Smoothing - Medium
WAN Smoothing - High
WAN Smoothing only needed if the environment having problem on VOIP or video streaming (packet drop, jitter).
Thank you for your suggestion, we will take into the consideration.
