GL-MT300A vs GL-MT300N-V2 OpenVPN performance


#1

I just wanted to ask real quick here, since I could not find anything specific regarding performance of the two CPUs (MTK 7620A @580Mhz and MTK 7628NN @580Mhz). Which one should be faster in regards to OpenVPN?

I currently have a GL-MT300A and am thinking about getting another one. With the MT300A and Private Internet Access VPN I usually get 8-9 Mbit/s down and up.

Is the MT300N-V2 faster or slower than the MT300A? On the product page “Faster OpenVPN encryption” is mentioned as a feature, so I’m wondering if that means anything (although I guess that “faster” here means faster than the original MT300N).

Also, would the AR300M still be the best choice when it comes to OpenVPN performance? Is it significantly faster than the MT300A?

I’m aware that questions like these get asked from time to time and there are a few answers, but never any information that compares all the devices.


#2

MT300N-V2 has 10% to 20% increase in vpn performance. It has hardware acceleration but we didn’t enable it. It is said to have 20% increase again.

In my test, here is a comparison of mini routers and vpn speed in real environment

MT300N 8Mbps

MT300N-V2 10Mbps

MT300A 13Mbps

AR300M 17Mbps

 

 


#3

Thank you very much! Is there a way for the user to enable the hardware acceleration for the MT300N-V2, or any plans to do that?


#4

The users cannot do that. It includes driver development. Because the vpn performance increase is not that much, we have to balance its real value with the development cost.


#5

I understand. So I guess the best way to go for performance is still the AR300M, correct? Will there be any new portable mini router in the near future that can do better than the AR300M? If so, I could just wait a bit to purchase a new one.


#6

Since I was curios and they’re pretty cheap, I decided to get an MT300N-V2 to test its VPN speed.

So through a wired connection (and using Private Internet Access as VPN provider) I get:

MT300A: 1.31 MB/s peak, ~10.5 Mbit/s

MT300N-V2: 1.34 MB/s peak, ~10.7 Mbit/s

Firmware version was 2.264 on both.

So surprisingly, the MT300N-V2 is at least as fast as the MT300A. Still slow, mind you - but now it would be really interesting to see how the performance would be if the hardware acceleration was being used.


#7

Is it possible to add this feature please?

 

It is a very good router, but OpenVPN still slow…


#8

now we checked more details about openvpn and it cannot use hardware acceleration because of its design.

ipsec maybe able to use hardware acceleration.

We are making another small router and it has more powerful CPU than AR300M and AR750. Hope it will be available one month or two


#9

What about a version with two wireless modules so it is possible to create an encrypted access point WLAN while connected to an unencrypted hotel WiFi? Also that way wireless performance would be much better.


#10

Just wait for wireguard VPN! Here are my results with clean LEDE on AR300M-Lite (iperf3 through wireguard tunnel between AR300M and a local test server):

root@LEDE:~# iperf3 -c 10.0.10.1
Connecting to host 10.0.10.1, port 5201
[ 4] local 10.0.10.3 port 50784 connected to 10.0.10.1 port 5201
[ ID] Interval Transfer Bandwidth Retr Cwnd
[ 4] 0.00-1.00 sec 6.43 MBytes 53.9 Mbits/sec 0 123 KBytes
[ 4] 1.00-2.00 sec 6.50 MBytes 54.5 Mbits/sec 0 158 KBytes
[ 4] 2.00-3.00 sec 6.56 MBytes 55.0 Mbits/sec 0 176 KBytes
[ 4] 3.00-4.00 sec 6.44 MBytes 54.1 Mbits/sec 0 186 KBytes
[ 4] 4.00-5.00 sec 6.50 MBytes 54.5 Mbits/sec 0 195 KBytes
[ 4] 5.00-6.00 sec 6.50 MBytes 54.5 Mbits/sec 0 206 KBytes
[ 4] 6.00-7.00 sec 6.62 MBytes 55.5 Mbits/sec 0 206 KBytes
[ 4] 7.00-8.00 sec 6.50 MBytes 54.5 Mbits/sec 0 206 KBytes
[ 4] 8.00-9.00 sec 6.81 MBytes 57.1 Mbits/sec 0 291 KBytes
[ 4] 9.00-10.00 sec 6.44 MBytes 54.0 Mbits/sec 0 291 KBytes


[ ID] Interval Transfer Bandwidth Retr
[ 4] 0.00-10.00 sec 65.3 MBytes 54.8 Mbits/sec 0 sender
[ 4] 0.00-10.00 sec 64.5 MBytes 54.1 Mbits/sec receiver

iperf Done.
root@LEDE:~# iperf3 -c 10.0.10.1 -R
Connecting to host 10.0.10.1, port 5201
Reverse mode, remote host 10.0.10.1 is sending
[ 4] local 10.0.10.3 port 50788 connected to 10.0.10.1 port 5201
[ ID] Interval Transfer Bandwidth
[ 4] 0.00-1.00 sec 7.79 MBytes 65.2 Mbits/sec
[ 4] 1.00-2.01 sec 8.71 MBytes 72.4 Mbits/sec
[ 4] 2.01-3.00 sec 8.31 MBytes 70.5 Mbits/sec
[ 4] 3.00-4.00 sec 8.69 MBytes 72.9 Mbits/sec
[ 4] 4.00-5.01 sec 8.68 MBytes 72.5 Mbits/sec
[ 4] 5.01-6.01 sec 8.44 MBytes 70.6 Mbits/sec
[ 4] 6.01-7.02 sec 8.78 MBytes 72.7 Mbits/sec
[ 4] 7.02-8.00 sec 8.34 MBytes 71.3 Mbits/sec
[ 4] 8.00-9.00 sec 8.60 MBytes 72.0 Mbits/sec
[ 4] 9.00-10.01 sec 8.44 MBytes 70.2 Mbits/sec


[ ID] Interval Transfer Bandwidth Retr
[ 4] 0.00-10.01 sec 86.3 MBytes 72.3 Mbits/sec 0 sender
[ 4] 0.00-10.01 sec 85.3 MBytes 71.4 Mbits/sec receiver

iperf Done.


#11

a new small router? can you provide any specifics?


#12

@Alzhao,

 

A new router sounds with more powerful CPU sounds great. I think what many people are looking for is hardware that has higher OpenVPN performance. Will the new router use hardware crypto rather than software which relies on a faster CPU?

 

While I love my AR300M I’ll pick up a new router immediately if it can improve on OpenVPN / Shadowsocks performance.


#13

Just a reminder: Hardware acceleration is not possible for OpenVPN, it’s only possible with IPsec. So for OpenVPN, it will always be up to the CPU to do the work.


#14

Yes. The results is, openwvn cannot use hardware acceleration, only possible for cpu power.


#15

Any link for your vpn protocol?

Also for local server is is much faster.


#16

Huh. After all these years I did not realize that. Seems crazy hardware acceleration is not used.

Alrighty then… more CPU power it is!


#17

I don’t think so. Why upgrade to more and more powerful hardware if performance can already be increased 10x-20x with Wireguard. I would hope for Wireguard Support within the VPN section of the next Firmware version. That would speed shit up a lot!


#18

That only works if the server uses it too, does it not? As long as OpenVPN needs to be used (because that’s the only thing a particular server offers), acceleration is just not possible. Then more CPU processing power is the only option.