Recommend me a VPN for SPEED

I’m shopping for a VPN service and the variety of options combined with the obvious schill-reviews all over the place are making it very difficult to get objective advice.

My #1 priority is speed. I have gigabit fiber to the home and I want to maintain the highest speeds possible.

Needs to run on Linux and possibly also run on my router in the future (I’m getting the Synology router soon), and needs to support torrents.

Is there any way for me to know what VPNs will be fastest without just taking the plunge and testing it myself?

Security / location of the company is unimportant to me.

How about sending them an email? They are businesses after all.

You… want me to send an email to a VPN provider asking them how fast they are?

do you… think there might be a possible conflict of interest there?

1 Like

Seeing as how that’s the single most important thing about a VPN, I’m not sure why you want to run in the first place.

1 Like

Roll your own using an AWS ec2 instance or gcloud compute instance?

On my 150 Mbps home connection, I get about 34 Mbps funneling my traffic through an AWS ec2 instance. And I get about 43 Mbps sending my traffic through a gcloud compute instance. Those are with un-optimized OpenVPN connections. The one to gcloud is even over TCP. The primary advantage of AWS is full ipv6 support.

1 Like

It’s the single most important thing to you, maybe.

The only thing I really care about is no longer getting wrist slaps from my ISP.

Considering the fact that all they really want is an excuse to turn a blind eye and keep my business, I think just about any VPN would do the job. And I would like to stay as close to my current speeds as possible in the process.

Sure, but it never hurts to ask. After all, they do also have an interest in providing their core service. They might be able to tell you whether their nodes close to you are of the recuired speed at all.

Also, I do know Mullvad (which I always recommend) have a 3 hour test period, so you could go ahead and test it. But I’m unsure whether you meant you don’t want to pay to test, or don’t want to test at all. .

i have pia and i did a 50gb download test with it and got 110mb/s average on my 100mb/s internet so i’d say it past.

1 Like

Well, if you care about a VPN actually doing its job to keep you safe and not just throwing money down a well with a performance hit then yes it is the most important thing to me.

Then you want a VPN that won’t buckle when your ISP asks your VPN provider for its logs about you and to see your traffic; which they can legally do depending on where the VPN is located geographically and on the VPN’s own policies about log retention (if any).

Long story short is that the best VPN you should get is one that:

  • doesn’t keep logs
  • not located in the US, China, Russia, anywhere where Big Brother can get in easily.
  • has servers close to you (Speed)
1 Like

I have not tried paid VPN’s to give a recommendation, but keep in mind there is no one metric for “speed”, so you probably need to think about exactly what you need.

For example any VPN will hurt latency, but by how much will vary.
Also most VPNs will have a bias towards download speeds, but if you care about upload that will rule out some options.
And location will matter since you will get a faster\lower latency service if they have end nodes near what your trying to access, however physical location also effects which laws they have to follow, so can get you stung by things like forced data logging\handover… the US based services being the most spoken about for this reason, but other countries having similar rules too.

I will Second PIA (Privateinternetaccess). Ive been using them for years and have never had a problem with throttling.

You can run your own VPN on digital ocean or another similar site.

1 Like

Thanks everyone for the suggestions.

I do like the idea of rolling my own, however at the moment I am looking for a ‘quick n dirty’ solution that gets the job done. I will probably pursue setting up a VPN of my own in digital ocean / aws / etc as a learning project in the future.

Given the wide variability of performance based on what ISP you use, your location, etc, I decided to simply bite the pullet and purchase a months worth of service from three providers - TorGuard, Mullvad, and PIA (based on the recommendations I received in this thread).

TorGuard does have a ‘premium 10gbit’ option for an additional $20/month, but that would be $30/month for my VPN - the speed might be great, but that’s just too much.

I went for the default package for each and ran speedtests on both Ookla and for each.

I got a pretty wide variety of results, sometimes even from the same provider. I also was not super careful about limiting traffic on my network when performing the test, so that may have played a role as well.

Mullvad performance was pretty bad - 53 down / 47 up / 22 ping in ookla, 16 down / 19 up / 24 ping in

Torguard was alright, but the performance seemed to vary a lot from test to test. I wonder if this is related to the fact that TG and PIA both use the ‘TAP’ driver for VPN. Each of them reinstall the driver during installation, but I didn’t reboot or anything in between. Anyway I got a result of 202 down / 167 up / 23 ping in Ookla, but only 12 down / 23 up / 25 ping in Very odd!

PIA gave the most consistent performance. I did get some weirder results after I installed TorGuard, but after removing TorGuard and reinstalling the TAP driver the speeds evened out again. I was able to get speeds of 205 down / 414 up / 22 ping in Ookla and 92 down / 23 up / 26 ping in I was even able to further boost this speed even further by disabling VPN encryption to 256 down / 412 up / 22 ping.

So, I’m sticking with PIA. Thanks for the suggestions everybody!


I did this with DO to learn how to do it and it does work well. Added bonus is you can snapshot it and spin it up in any on the data centres worldwide on demand and turn it off when not needed. However the $5 droplet only has 1TB/month.

1 Like

Not at all. If you want to test actual speeds instead of your ‘local’ network have a look at

I used to run droplets at DO for over 2-3 years, and during that time, I’ve had ‘hardware’ issues meaning I’ve had to migrate stuff at some point or the other. This is OK if you have Ansible playbooks for everything and take a DevOps approach - but this isn’t the case for everyone.

In the same span, I’ve had little to no issues with EC2 instances. YMMV of course, but in terms of ‘stability’ I’ve personally seen less issues in AWS.