LTE Speedfusion Fail

I’ve been working with Speedfusion experts, Pepwave technical support, and 5G technical support for awhile now.

It is becoming apparent that Speedfusion bonding of LTE (only) WANs simply does not work, despite the marketing information that promotes improved bandwidth of LTE connections. No, it does not only promote reliable VPN, always up connections; it also promotes improved bandwidth.

My Speedfusion bonded LTE connection tests at significantly less than HALF of the combined speed of the individual LTE connections over the Max Transit Duo. Additionally, the individual LTE connections tested locally on my Max Transit Duo don’t achieve half of what they do on my Galaxy s10+ phone (ATT) or my Jetpack 8800L (Verizon). These devices are using the same bands as the Max Transit Duo so this is not the issue. I understand the difference between cat 12 and cat 18 but the disparity in speed doesn’t begin to be explained by the modem technology.

I am not using the Max Transit cellular paddle antennas. They are even worse. I am using a roof mounted Mobilemark 5in1 MIMO antenna.

It is about time that Peplink get serious about the problems with the Speedfusion solution with LTE WANs. I have not found ANYONE that has had success with these technologies with LTE.

Either Peplink should change their marketing or publish obvious disclaimers about Speedfusion and LTE. It is misleading.

PROVE ME WRONG: HAS ANYONE HAD SUCCESS WITH SIGNIFICANTLY IMPROVED BANDWIDTH USING SPEEDFUSION WITH ONLY LTE WANS? Not only is it not improved, Speedfusion doesn’t even give bandwidth of either individual, non-bonded connection, and these are about half of what other devices give.

Something should be escalated in my humble opinion. These are not inexpensive solutions.

Can you please share what each individual LTE WAN is getting in terms of speed? Are they consistently getting those speeds? Are you sure when you are comparing the device to your phones, even though they are on the same bands, they are connecting to the same cell tower?

1 Like

Hi Joel,

We had plenty of good experiences with speedfusion bonding.
Stability and throughput wise Could you tell me how you did the test?

  • Providers used in the same bonding test
  • Freq used
  • Units used (concentrator & bonder)
  • LTE, LTEA or 5G tested

Also keep in mind you can never go higher when bonding then the specified Pepvpn throughput.

2 Likes
  • Peplink Max Transit Duo w/Mobilemark 5in1 MIMO antenna on roof
  • Speedfusion hosted on Solo at Vultr (closest Vultr site to testing location)
  • AT&T & Verizon non-throttled, non-network managed
  • LTE and LTEA - tested using default (Auto) connections as well as each and every selectable band
  • Tested at all times of tower congestion - none to very congested
  • Tried every adjustment possible for latency management
  • Compared Max Transit Duo to other LTE devices (using same bands)
  • Tested using WAN testing and Speedfusion testing on both local Max Transit Duo and Fusionhub ends of the tunnel
  • Also tested using client speed testing apps; many, and always used testing sites close to Vultr

Never had a bonded test that equaled any one of the individual LTE WANs.

Please tell me about one of your LTE-only implementations. Thanks.

Hi Joel,

Fusionhub solo will only allow 100MBPS of MAX troughput. But you prob know this.

Test we Did with LTE in the Netherlands.

Peplink BPL-380 1GBPS Fiber U/D (concentrator)

Max-HD2-LTE-E-T (bonder) Puck-2 antenna 2x
LTE1 1800MHZ B3 Tmobile
LTE2 1800MHZ B3 Vodafone
image
Over Single LTE we got on both lines arround 60-70MBPS down and 30 UP
When bonding we managed to push the unit (unencrypted) to 118MBPS Down and 58 UP

Above test is just a simple example LTEA + 5G requires extra setup. Also taking into account multiple factors.

The HD4 MBX we managed to get 500/180, but coudnt max it out yet due to no 5G enabled infra here.
Test providers: Vodafone / KPN / Tmobile Combined.

2 Likes

FusionHub solo doesn’t have 100Mbps limit anymore:

Q6. Does FusionHub Solo have any bandwidth limitation?
There is no bandwidth limitation.

From FAQ on Peplink.com.

3 Likes

I think what you need to keep in mind is that bonding isn’t designed to “double” your speed for example, but it’s there to provide better reliability and performance. By using multiple SIM cards you can reduce dropped packets from just 1 SIM card alone.

If you’re using 2 SIM cards from the same provider then that can be the issue also since each cell tower only has so much bandwidth available to it so it’s better to mix multiple providers. You’ll also always get way faster speeds on a consumer cell phone like the S10+ as it uses the latest radios as is designed for single user use. The Peplink products are designed to support hundreds of users on them, where as you could never have that many users all running off the CPU in the S10+.

The most important thing to remember is that the primary use of bonding is NOT to give you more bandwidth, but is to give you better stability and reliability.

3 Likes

Thank you Wouter. I have been testing with different Latency Difference Cuttoff (although never this low) AND Cut-off latency (on the Fusiionhub side). I haven’t achieved this kind of relative performance though.

I’m moving to a different location with AT&T and Verizon towers that are in my back yard and never very congested. I’ll keep configuring and update my results in about a week.

Thank you VERY much.
Joel

Hello Lee_Moreau. I’m using AT&T and Verizon (on different towers). I understand about the Galaxy improved (cat18) modem. Bonding should have a, best case, 17.5% degradation but as you can see from my results, this doesn’t come close to explaining my bonding experience. I’m getting bonded results that aren’t even equal to either of the LTE WANs individually.

Thanks for your input, I will take all that you’ve said into account in my testing and report my results going forward.

Cheers,
Joel

Missed that one seems to be changed indeed thnx for the update

2 Likes

Hi Joel,

You are welcome, cutoff latency is only needed on the LTE side of the tunnel.

Hope you will achieve some results that leave you more satisfied.

Gr,
Wouter

1 Like

Hello Lee. The Max Transit Duo isn’t marketed to support hundreds of users. It doesn’t have the chops to do it. But, even with a few users, bonding could be important. Wishful thinking about improved performance as the marketing indicates.

It’s interesting that I experienced the exact same behavior using the Sharedband service, except I was using identical WANs from the same carrier. The bonded service generally delivered half the speed as each individual WAN. They did a lot of hand waving and excuse making before they stopped providing service in the US. So now I have an expensive proprierary router that is useless to me. I’m hoping to not go down that road again.

It’s interesting that I experienced the exact same behavior using the Sharedband service, except I was using identical WANs from the same carrier.

Were you using one of the major US cell carriers, or a different carrier?

1 Like

I’m using one Verizon sim and one AT&T sim about equal in RSRP, RSRQ, and SINR. Both have good bandwidth and a few lost packets.

1 Like

I was using two sims through AT&T.

1 Like

We have an open ticket on this and Verizon connects with faster LTE-A service but the signal quality is bad.

The auxiliary antenna showed poor quality (around -128dBm) for RSRP as compared to the main. SINR was poor (around 0dB) on the main however, and both antennas showed a fair to poor RSRQ value (around -15dB and -13dB) in our testing.

Additionally our tests showed bonded throughput was much higher than a single WAN but it was not consistent due to poor signal quality.

1 Like

This open ticket is not mine. The numbers and results (all of them) are not representative of my connections- not even close. What ticket are you referring to?

Ticket number is 20020218 and screen shots were provided. If this is not your ticket please open one with us here.

1 Like

Dear All,

I have the same issue here:

  • two LTE connections
  • different performance
  • the result is only the avarge speed of both not the total of both ( minus overhead)
  • what is strange that upload seems to be limited to 50 MBps

See screenshot. I hope it is self-explaining:

Strange that there are so many abadonned threads on this. I hope someone has a solution idea.