I am using MAX-OTG-U4-SF installed in a car. The unit is equipped with 3 USB sticks. Two of them are 3G and one is LTE. On the other side the Pep VPN is terminated with BALANCE 380 in a data center with a direct link to the mobile operators (speed is above 70 Mbits and latency is around 4ms).
When I roam around the city I do regular speed tests and get unsatisfactory results (sometimes below 0,25 Mbits). I have attached a snapshot from the Pep VPN analyzer and it is obvious that the combined speed of the 3 connections is much lower than the combination of two of them or even a single connection.
I have done the test with 3G sticks only and achieved simmilar results.
Did you enable your outbound policy settings using the PepVPN as the first priority?
Also, the way I understand it is Peplinks bonding is not truly additive. If you bond a slower connection with a faster connection, the speed will be somewhere in the middle, not the combined bandwidth of both. I believe the same goes for the Latency.
It could be like that. In that case Speedfusion is not very usefull for 3G/4G connections, because when you move around you rarely loose any of the connections. More often some of the lines get slower when their signal is weak. The more 3G/4G connections you use (3 in my case) - the higher the chance is for at least one of them to be on a weak signal. And if bonding slower with faster connection results in speed in the middle, you experience speed degradation instead of speed fusion as seen in the example I placed above.
I hear you man, I am havjng my own issues and am getting pretty frustrated. The only helpful tech support I’ve gotten so far has been from other members of this forum.
Try this:
Change your outbound policy to force your traffic through the speed fusion vpn
Run multiple simultaneous downloads through your connection and watch your overall throughput.
I have deployed multiple cellular bonding solutions from different vendors, and all of them have issues delivering 100% of the bonded bandwidth to a single user. Also, bonding seems to increase the stability of Cellular links in most of the systems I have used.
Have you made any progress with this? I recently got a MAX OTG U4 and have not been able to achieve any increase in bandwidth from bonding. I am running the latest firmware and have “All Traffic Send To” set to go through the SpeedFusion vpn. I believe this is equivalent to what TDComm refers to as “force your traffic through the speed fusion vpn”
Is there anyone out there using MAX OTG U4 with 4G devices and actually getting faster speeds than that of the fastest 4G device used independently of the MAX OTG?
The support team was following my case untill they achieved some good test results while the unit was stationary.
My experience is that speedfusion works only on connections with simmilar quality - throughput and ping. If some of the connections is with lower quality it degrades drastically the performance of the Pep VPN. Unfortunately I want to use the unit while travelling in my car and in that case there is always one connection with bad quality that affects the other. Even worse, the more lines you use, the higher the chance for bad connection on one of them, so you have 100% speed degradation guarantee with the current Speddfusion algorithm.
I’m trying the new beta release that has some options for cut-off latency and packet lost pull-back time but they simply don’t work.
Best luck and share your good experience if you have any.
As I understand it, your outbound policy needs to looks this this in order for SpeedFusion to bond your traffic.
Here are the algorithms you can currently use for directing your WAN Traffic:
Weighted Balance - Traffic will be proportionally distributed among available WAN connections according to the specified load distribution weight.
Persistence - Traffic coming from the same machine will be persistently routed through the same WAN connection.
Enforced - Traffic will be routed through the specified WAN connection regardless of the connection’s health status.
Priority - Traffic will be routed through the healthy WAN connection that has the highest priority.
Overflow - Traffic will be routed through the healthy WAN connection that has the highest priority and is not in full load. When this connection gets saturated, new sessions will be routed to the next healthy WAN connection that is not in full load.What I would love to see if a fail over algorithm that picks the modem with the best throughput/latency.
What I would love to see if a fail over algorithm that picks the modem with the best throughput/latency.
I’ve used some really high end bonding software/hardware that they use to stream live video feeds for Television Networks and such, we still had the problem Platts is describing. An extremely low throughput connection drags the rest of the connections down.
If you want to go to the high end market you’re going to be spending upwards of $2,500 for a single unit. The bonding works quite a bit better, but it still has limitations.
In my opinion Peplinks real competitor is Cradlepoint, not Mushroom Networks or TeraDek.
The way I see it, for mobile applications, bonding isn’t the way to go due to the problems discusses above.
Let’s think about some options, maybe we can trick the unit into doing what we want.
The unit could be setup in a failover mode, so when the signal/throughput from one modem/carrier degrades, it switches over to the next modem/carrier. In this way, you are always connected to the best modem/carrier.
So, how can we convince the Pelink that the modem needs to turn off when the throughput/signal is weak?
Could you change the settings on the health check method to accomplish this?
Would that make your modems sensitive enough to a lack of throughput/signal for it to fail its health check?
I noticed that in the new beta firmware they have introduced lowest latency as outbound algorithm for all devices including MOTG. Tomorrow I will disable the PepVPN and give it a try to see if it will switch to the fastest line while driving around the city. I guess this is the option you are looking for.
Hi, I see your points, and what Platts suggested here is a good thing to try. More information about “Cut-off latency” in this post (as Platts already replied): Peplink | Pepwave - Forum
While please note that you need to have both PepVPN peers upgraded to 6.2.0 beta 3 and configure this Cut-off latency settings on both ends.
Hi Platts, one thing I’d like to explain more about is the PepVPN Analyzer, this tool only test the performance on transmit direction, if you want to get the performance data on receive direction, you’ll need to run PepVPN Analyzer on the remote side (data sent from remote > local, and thus testing the local receive direction).
Actually bonding 3G and LTE is not always a good idea because the overall latency will be degraded by the 3G connection, and LTE can be practically almost 10x the throughput of 3G (e.g. LTE give you 20Mbps DL and adding up 3G with 3Mbps DL doesn’t make a big difference and will even degrade the performance because of the large latency of 3G network), also with the overhead introduced by PepVPN, you may not always get the expected bonding result. The best usage with 2x3G and 1xLTE is using LTE as the 1st priority connection, and use the two 3G as 2nd priority for Hot Failover to form an unbreakable VPN.
That’s the problem Steve. Here in the US we roam from 3G to 4G as you move around the country. We would like to be able to use a 4G connection whenever possible, but the Peplink cannot distinguish between a 4G, 3G or 2G network. So, what ends up happening is whatever card is placed in WAN 1 soaks up the majority of the traffic, wether or not WAN 1 is indeed the fastest.
Example:
WAN 1: Verizon 3G
WAN 2: AT&T LTE
In the above example, if it wasn’t for the Verizon modem I would be able to fully utilize the AT&T LTE connection for my bandwidth. Bonding doesn’t help me because the increased latency and degraded throughput degrades my 4G connection.
You guys need to have a algorithm that picks connections based on throughput, or at least signal technology. The device already knows if the connection is 4G, 3G or 2G, how hard would it be to have the Peplink prioritize connections based upon which signal technology the modem is currently using?
I understand that you guys have your support staff overseas, and maybe even your product development team. So, I want you guys to fully grasp the geographic vastness of North America and how many areas are currently served by older signal technologies. Our modems constantly flip between 3G and 4G as we travel between cities, you guys need a way to manage this.
Today I tried to disable the PepVPN and used the lowest latency outbound rule (firmware 6.2.0 beta3). At first the results were great due to the fact that MOTG compared all lines and picked the fastest one (which is the idea behind the lowest latency outbound rule). Then while driving around the problems started to show up. It seemed as if MOTG was stuck with the first choice line and didn’t change it untill it was totally off.
It will be helpful to know (and if possible to adjust) the time interval between the individual latency checks that are used to compare the lines. Obviously when you move you need shorter intervals between the latency checks in order to stay on the best connection. It seems to me that MOTG makes these checks every minute which is quite long time while driving. I think this will resolve the above issue described by TDComm as well.
Believe you are testing with TCP connection. TCP is a Connection-Oriented protocol. It is having connection timeout. This is defined in RFC standard and should follow by network related product. Thus your connection still “stuck” in WAN1 (for example) before the timeout. This is normal behavior for TCP connection.
You are testing the exact issues that I am. I am waiting on Peplink engineering since Monday to give me feedback on my setup so I don’t want to make any changes in the event they are. Beta 3 seems to have features that may improve the performance. I want to understand want you tested. “disable the PepVPN” means you are not using the bonding capabilities of the device? A successful test for me would be an upload speed that is greater than the speed of the fastest 4G device attached, did you achieve this?
I don’t belive you will have much success unless all your lines are of simmilar quality. Most probably your throughput will be the result of the slowest line multiplied by the number of lines you are using.
Chech this thread for explanation: Peplink | Pepwave - Forum
Have you observe what is the latency when particular 3G/4G connection getting worst when you roam around the city? You may record these figures which can use for Cut-off Latency in SpeedFusion profile. This will help to suspend the WAN link temporary which having high latency.
