WAN Smoothing vs Bonding

Dear Guys,

Can someone fully explain what is the exact difference between SpeedFusion WAN smoothing offered by MAX On The Go and SpeedFusion Bonding offered by MAX-700 ? we could not fully understand these functions as on both of them all WAN links are active on the SpeedFusion tunnel.

Thanks a bunch

Hi Hootan,

SpeedFusion WAN Smoothing and SpeedFusion Bonding can’t really be compared.
When you use WAN smoothing you use WAN bonding as well; WAN Smoothing is a feature that can be enabled for the SpeedFusion tunnel.
.
SpeedFusion bonding, truly bonds several WANs on a packet level, which is explained here.

All WAN’s can be combined to form one virtual WAN; packets at the local site are sent through each of the WANs and reassembled in the original order at the remote site of the SpeedFusion tunnel.

WAN Smoothing is a different feature that can be enabled for SpeedFusion.
Instead of using 1 packet WAN Smoothing will send packets over 2 or more WAN connections and only the packets that arrive first will be used (other packets get discarded)
Thus, the latency of the SpeedFusion tunnel becomes the latency of the most responsive WAN connection.
This is extremely useful for deployments where improving consistency is more important than improving bandwidth.
WAN Smoothing is only needed if the environment is having problems on for example VOIP or video streaming (packet drop, jitter).
More information on WAN smoothing can be found here.

3 Likes

As Erik says. SpeedFusion Bonding is packet level aggregation of traffic over multiple WAN links.

WAN smoothing is a feature enhancement of Bonding where sent packets are duplicated across all wans to mitigate the effect of those WAN links that have packet loss or variable latency.

You can think of WAN smoothing for WAN links as being similar to RAID for disk arrays.

3 Likes

Dear Martin and Erik,

Thank you guys for the complete explanation of these two features.

Have a great day.

1 Like

I am confused. You said that when you use Smoothing you use Bonding as well but then why do some products offer Smoothing and not bonding too? I.e. the Peplink Max BR1 line.

Is a special configuration, VPN, device, anything needed to take advantage of Smoothing or Bonding? You mention a “tunnel” is needed to use Smoothing? So I can’t just connect a few WANs to one device (i.e. MK2 or 210/310) and have that one device provide connectivity to vehicle/RV/boat? What is this tunnel I need?

My use case is as follows: installation in a vehicle, RV, or boat… I will have 2-3 WANs: a) cellular, b) WiFi as WAN 2.4ghz, and c) possibly WiFi as WAN 5ghz.

In some cases 1-2 of the WANs may be unreliable especially as a result of movement but I do not want to have video conference, VoIP, etc drop. For that it sounds like WAN smoothing will help?

What is the maximum speed if speedfusion is enabled (smoothing or bonding)? Will using bonding (not available on all peplink products) instead of smoothing provide a faster connection in times when one WAN is significantly slower than the other? Or is Smoothing smart enough to handle both reliability (no connections dropped) AND speed (multiple WANs working together)?

Now to complicate things further, what happens when one WAN has a lower latency but is slower? For example, I am finding WiFi to have a lower latency than cellular but bandwidth seems to be significantly lower at times (i.e. Cellular might be 30mbps down / 5mbps up with a higher latency and WiFi may be 5 mbps down / 1mbps up with a lower latency). With Smoothing does that mean I am limited to the lower latency WAN and thus a max of 5 mbps / 1mbps speeds? Or can it somehow combine the WANs and I will have 35mbps/6mbps available? What happens if I use Bonding and one connection is bad/very high latency/packet loss/etc, how will Speedfusion handle?

I am just trying to figure out which device I should go with and which Speedfusion technology I would use/configure?

For example, a MK2 has cellular and dual-band simultaneous WiFi as WAN available but it says it can only do Smoothing/Hot failover not Bonding. Am I better off getting a Router such as the 210/310 that can do all three: Smoothing, Bonding, and Hot failover and then have separate devices connected to it (cellular router/modem, WiFi AP for 2.4GHz, and WiFi AP for 5GHz)? I’d much rather manage everything from one interface instead of having to log-into four separate devices though… (peplink router + those three WAN devices).

Maybe I am making this more complicated than it needs to be?

Thanks for any insight!

No, I said WAN smoothing is a feature enhancement of bonding - which perhaps isn’t accurate. More accurately WAN smoothing is a feature enhancement of the SpeedFusion Technology (of which bonding is one feature).

Bonding bandwidth across multiple WAN links has a considerable processing overhead - especially if encryption is enabled. Smaller devices like the BR1 would not cope very well with bonding WANs and overall throughput would be degraded. Duplicating the traffic and sending it across multiple WAN links in a fire and forget kinda way requires less processing power.

All SpeedFusion technologies are point to point ie between one Peplink appliance and another. So to take advantage of Smoothing or Bonding you would need another appliance physically hosted in a datacenter or virtually hosted in the cloud that you could create a SpeedFusion connection to. Then you would be able to use bonding or wan smoothing if your device supports that.

WAN smoothing would help in this case yes.

WAN smoothing replicates traffic across multiple WANs. At the lowest setting this is a duplication of the traffic that is sent over multiple WANs. At the very least then you will half the total available bandwidth to you if you turned WAN smoothing on for all client traffic.

Speedfusion Bonding aggregates the available bandwidth. In ideal conditions you will be able to use around 80% of the total available bandwidth as 20% is used by SpeedFusion to manage the VPN. However if there is packet loss and jitter on individual WANs, the latency over the tunnel will vary considerably and available bandwidth / throughput will vary too as SpeedFusion looks to manage how those WAN links are used. Since this is also a point to point technology - the latency between you and the remote data center / cloud is also a factor.

With WAN smoothing data will be duplicated across the multiple WAN links. If you were in a video call that was 2MBps/ bidirectional the WIFI connection (with its 1MBps up) would saturate very quickly but the video would work fine over the cellular connection. The device at the remote end would just discard all the packets coming in over the wifi connection as they would be late due to increased latency.

You will have in the region of 24/4Mbps available to you when using WAN smoothing as SpeedFusion has a 20% ish overhead and you are duplicating the traffic over WIFI and Cellular.
So a VoIP call @ 160Kbps would be sent and received over both the cellular and the wifi connections at the same time. If one of those connections failed completely, had a massive latency spike or was throttled then you wouldn’t notice as the packets would still get through on the other connection.

Buy a BR1 MK2. Only turn on WAN smoothing for real time traffic (VoIP / Skype etc) so you are only duplicating the most time sensitive traffic, and just use load balancing for everything else so you get to use all the available bandwidth. Host a free virtual FusionHub Solo appliance in Vultr.com for $5/month as the central end point. Job done.

2 Likes

Thank you for your insight.

All of my confusion started after watching the 3gstore video at: https://www.peplink.com/technology/speedfusion-bonding-technology/

I dont think there is any mention of having to have a second Speedfusion enabled device or appliance setup in a data center at any time during the video. The way the setup is described is everything that is happening is as a result of equipment right there in the video. So now I am a bit disappointed to learn that in order to achieve the result of what we see in the video I would actually need a second device or to host a virtual FusionHub Solo appliance.

In any event, $5/month is not bad and perhaps its something worth doing. Is there a guide on which Vultr package to select, how to install, configure, etc? Would I be setting up a VPN? I just don’t know how I feel about routing all of my traffic to an extra place…and if the provider/data center goes down that may mean that I have no internet? What if I don’t have or use InControl? Or is that a pre-requisite?

So if Smoothing and Bonding do need a second device or appliance located/hosted elsewhere, what about Hot failover? Does that feature require the same and cannot be used independently on a single device?

Thank you very much

Not during the video no, but the diagrams below show two devices with SpeedFusion between them. The requirement for two appliances regularly causes confusion.

No. I really ought to put one together.

Fundamentally yes, a SpeedFusion VPN tunnel from the onboard device to the hosted appliance.

Unlikely to happen - I have servers on Vultr that have been up for over three years with no outages. If it does you can set any traffic that normally uses SpeedFusion to break out locally instead if the VPN is not available.

You need to use InControl2 to generate the license for FusionHub Solo. There are more reasons to use IC2 than not in my opinion, but you could just create an account to generate a license and then not use IC2 after that if you like.

All the really clever stuff requires SpeedFusion which requires another appliance hosted somewhere else. Load balancing, session based fail over of traffic between WANs, WAN prioritization - these approaches do not required SpeedFusion.

1 Like

Thank you. You have cleared up a lot of confusion. I actually watched the video elsewhere (on Youtube) so there was not even that diagram or any explanation with it, I was just going off the video.

Regarding the failover, my Balance 20 shows “cold” standby…can I do “hot” standby with a 210/310/MK2/etc? Is hot standby the same thing as “Speedfusion hot failover”? Can you please clarify if the Fusionhub or a second device/appliance is needed for “hot standby” or just “speedfusion hot failover” and what the differences between those are?

Sure. Hot standby is WAN availability term perhaps best demonstrated with LTE connections.
Lets imagine you have a wired WAN (a DSL or cable modem) and an LTE connection.

If you set the LTE connection to cold standby it is turned off and disconnected when in standby. This means that if the higher priority WAN link were to fail, the Peplink appliance would firstly enable the LTE connection - causing the module to initialise, read from the SIM then start the potentially lengthy process of trying to find and connect to a cell tower. Only once the LTE WAN had connected to the cell tower, got an IP address from the provider and then passed its health checks would failover complete and user traffic from the LAN pass over the LTE WAN.

If that same LTE connection is in hot standby, then even though it is in standby mode the LTE module is live and there is an active connection to the internet over the mobile operators network but the Peplink appliance is not actively using the LTE WAN for traffic until it detects that the higher priority WAN link fails.

When the higher priority WAN link fails, because the LTE WAN is ready to receive traffic (it has an IP and is connected to the MNO), although all active sessions are ended (because the higher priority WAN link disconnected) they can all be rebuilt very quickly over the LTE WAN. In these kinds of failover situations most internet traffic will recover very quickly indeed, so if you’re browsing the web you generally wouldn’t notice a failover event. However if you were on a VoIP or VIdeo call at the time the WAN failed your call would typically be dropped and you would have to redial.

SpeedFusion HotFailover has the same concept as hot standby above but takes it a step further. Because SpeedFusion can control data flowing over the WAN links at a packet level (rather than just the session level that you’d see with typical WAN failover), when the primary WAN link fails, packets within the live sessions that were flowing over that WAN link are immediately redirected over the hot standby link. This keeps the real time traffic sessions up so the VoIP and Video calls do not drop.

So in summary. SpeedFusion Hot Failover is a SpeedFusion technology and requires a P2P Speed Fusion VPN connection to another appliance to work. Hot and Cold Standby can be found on all multi-WAN Peplink products.

2 Likes

Thank you, you are extremely helpful, your time is much appreciated. Now I think I understand everything but I am trying to find hot standby on my Balance 20 HW6 and cannot find that option. I do see “cold standby” but no idea how to change it to “hot standby”? I presently have a Mikrotik AP serving WiFi as WAN and a Cradlepoint 850 serving AT&T cellular as WAN.

I at least want to test to see if this more simple setup might fit my needs before diving into a 210/310/305/MK2 and setting up a Fushionhub.

One of the big issues with the “cold standby” was I could not access the router/modem of the WAN that was in “cold standby” for instance if I needed to tweak a setting or fix the connection I could not do so.

Go to the WAN settings (Network > WAN) and click the WAN that is in cold standby.
Change the connection priority to Always On. Click Save and Apply.

Your B20 is now load balancing traffic across the available active WANs.

To have hot standby between WANs instead, create an outbound policy of Any to Any, Type of Priority and drop the WANs into the desired failover order, then click Save and apply.

1 Like