5 replies to this topic

[dwright]

I already use port forwarding for torrents and get speeds that I consider very good, the fastest I've seen being about 4mb/s. However, when downloading things with Firefox it's a different story. On speedtest.net, I score about 4mbps, though this is bits rather than bytes if I understand rightly, meaning about 400kb/s second, as opposed to about 60mbps or 6mb/s with no VPN. Please correct me if I'm wrong with these conversions!

 

In any case, I had to download some large files with my browser and resorted to doing it with no anonymity software because the speeds were so much better.

 

Would I be able to mitigate this by forwarding ports for Firefox? Is such a setup possible?

[dwright]

Sorry to be a pest, but I thought I'd bump this thread since I've just experimented with the new speed test and speedtest.net.

 

I saw one post saying that you can get different speeds with different ports, so I thought I'd try Carinae on all the available ports: TCP 2018, 443, 53, 80; and UDP 2018, 443, 53, and 80. The following results are presented in that order.

 

Speedtest.net (download/upload):

 

9.18/2.41

9.72/2.20

13.59/2.73

9.39/2.23

4.93/6.90

3.39/7.93

2.65/8.31

4.19/8.10

[without VPN] 55.92/18.00

 

AirVPN speedtest:

 

Down: 8.380 Mbit/s Out, 6.908 Mbit/s In (82%), 20MB - Up: 14.230 Mbit/s Out, 2.587 Mbit/s In (18%), 20MB - Date: Thu, 17 Apr 2014 16:06:37 GMT - Buffers: 20MB/20MB - Laps: 3, Time: 179.93 secs
Down: 8.330 Mbit/s Out, 7.469 Mbit/s In (89%), 20MB - Up: 14.273 Mbit/s Out, 2.661 Mbit/s In (18%), 20MB - Date: Thu, 17 Apr 2014 15:58:39 GMT - Buffers: 20MB/20MB - Laps: 3, Time: 174.99 secs
Down: 8.110 Mbit/s Out, 7.752 Mbit/s In (95%), 20MB - Up: 14.376 Mbit/s Out, 2.729 Mbit/s In (18%), 20MB - Date: Thu, 17 Apr 2014 16:02:55 GMT - Buffers: 20MB/20MB - Laps: 3, Time: 172.78 secs
Down: 8.218 Mbit/s Out, 8.190 Mbit/s In (99%), 20MB - Up: 14.349 Mbit/s Out, 2.712 Mbit/s In (18%), 20MB - Date: Thu, 17 Apr 2014 16:10:44 GMT - Buffers: 20MB/20MB - Laps: 3, Time: 170.45 secs
Down: 8.284 Mbit/s Out, 4.621 Mbit/s In (55%), 20MB - Up: 13.972 Mbit/s Out, 10.950 Mbit/s In (78%), 20MB - Date: Thu, 17 Apr 2014 16:14:19 GMT - Buffers: 20MB/20MB - Laps: 3, Time: 125.57 secs
Down: 7.856 Mbit/s Out, 4.023 Mbit/s In (51%), 20MB - Up: 14.328 Mbit/s Out, 10.059 Mbit/s In (70%), 20MB - Date: Thu, 17 Apr 2014 16:17:34 GMT - Buffers: 20MB/20MB - Laps: 3, Time: 139.09 secs
Down: 8.379 Mbit/s Out, 4.703 Mbit/s In (56%), 20MB - Up: 14.074 Mbit/s Out, 12.424 Mbit/s In (88%), 20MB - Date: Thu, 17 Apr 2014 16:20:34 GMT - Buffers: 20MB/20MB - Laps: 3, Time: 121.14 secs
Down: 8.214 Mbit/s Out, 6.092 Mbit/s In (74%), 20MB - Up: 15.510 Mbit/s Out, 14.304 Mbit/s In (92%), 20MB - Date: Thu, 17 Apr 2014 16:23:36 GMT - Buffers: 20MB/20MB - Laps: 3, Time: 106.37 secs
[without VPN] Down: 8.190 Mbit/s Out, 6.315 Mbit/s In (77%), 20MB - Up: 15.238 Mbit/s Out, 15.398 Mbit/s In (101%), 20MB - Date: Thu, 17 Apr 2014 16:26:02 GMT - Buffers: 20MB/20MB - Laps: 3, Time: 114.16 secs

 

What I find interesting is A) that speedtest.net shows on the whole higher speeds, and B ) that both tests show decreased download speeds and increased upload speeds on UDP ports.

 

Having done this, no single port seems best for me (the score of 13.59 being anomalous it seems - I repeated the test for that one and it was unexceptional, and seems to be so on Air's test as well). Maybe I would like to use TCP more in the future though.

 

Still interested in the possibility of port forwarding more though. Is it possible to route my browser's traffic through a forwarded port?

 

[EDIT] I re-tried the test without VPN on Air's speedtest because I thought it must be an error - it was, as it said I must be connected to a server to perform the test.

 

As you can see from the speedtest.net result, my connection is quite a lot faster without the VPN.

[Staff]

Hello!

A remotely forwarded port is necessary when a service needs to be reached from the Internet. This is not the case of Firefox or any other web browser.

Units:

8 Mbit/s ("megabit per second") = 1 MB/s ("megabyte per second") = 1024 kB/s ("kilobyte per second").

That said, your system in-tunnel performance appears to be too low. Given that you have already tried all ports and modes on several servers, check whether your system has some non-standard network manager (such as Asus SmartConnect and similar) because some of them have been reported as dramatically slowing down tun adapter throughput (the tun/tap adapter is the virtual network card used by OpenVPN).

Additionally, please make sure that either on your router or on your computer no packets inspection tools are active (inspection of encrypted packets is not only useless but can slow down your system).

 

Last but probably most importantly, do not miss the following post:

https://airvpn.org/topic/11067-how-to-improve-average-speed/#entry15415

 

Kind regards

AirVPN Support Team

[NaDre]

I am not an expert on these issues, but I know that there a couple of things that make comparisons quite difficult.

 

1) There will certainly be increased latency when running a speed test, and because speed tests will involve some back and forth handshaking, they may produce a result that look substantially lower than what you would get if you were just downloading.

 

2) "Peering" arrangements between IPS-s and back bone exchanges can have a huge impact. Struggles over revenue sharing between businesses (particularly in the U.S. I believe) results in deliberate throttling. When you go through a VPN, the series of peer to peer exchanges you go through may change dramatically. And can impact bandwidth dramatically. So changing servers can make a difference. So can changing ISP-s. But predicting ahead of time what might help, unless you want to become an expert on internet peering issues, is impractical I think.

 

UPDATE:

 

What I know about these peering issues comes from the struggles my seed box providers has had providing fast connections to there users. On of the things they often ask for is a tracert/traceroute traces from the the users system to theirs, but also from their system back to the user's system.

 

The traceroute back to the user's system can be done from a shell account on the server. Something VPN users do not have.

 

These logs can be used by those who really want to dig into it to figure out what peer to peer exchanges are used going each way.

 

Given that there are so many technically oriented/curious people using AirVPN, I wonder if AirVPN would may want to add a feature to allow a user to obtain a traceroute back to their system from the VPN server?

[dwright]

Thanks for the helpful responses!

 

I've looked through the different possible causes. I don't believe I have non-standard connection software installed. My wireless card is "Intel® Centrino® Wireless-N 2230" - thought I'd say that, in case it helps you to suggest something to look out for.

 

I don't think there's packet inspection going on, though it's possible. I hope not. I use Malwarebytes Pro and Spybot S&D on Windows 7 - to my knowledge these don't perform that, but if you know otherwise, please correct me.

 

I think I'll try the mssfix mentioned in the other thread but I don't know how to implement that. Do I do this from the config generator? (I looked but can't see it)

 

EDIT: Wait, unless Intel PROSet/Wireless Tools counts as a non-standard network manager?

[dwright]

I tried uninstalling the Intel wireless programs and succeed in breaking my wifi lol. However all was well after a system restore.

 

I think the next thing I'll try is the mssfix but I'm still clueless about how to actually do that. Please help?