Change base transfer protocol (TCP/UDP) in VyprVPN?

Greetings!

I have been having some persistent speed issues with VyprVPN. To help solve this I decided to do some testing.

I did clean installs of VyprVPN and ExpressVPN. (NOT at the same time) I uninstalled, cleaned up, and cleared the Registry of residual entries for each VPN during the test any time I switched them.

I tested both using OpenVPN/256 bit

What I discovered:

  1. VyprVPN was noticeably slower than ExpressVPN.
    Download speeds were cut in half and upload speeds were almost a full order of magnitude (1/10th) the speed without a VPN.

  2. ExpressVPN speeds varied between five and ten percent slower than my “open” (no vpn) speeds.

  3. I discovered that ExpressVPN defaulted to the UDP transport protocol, so I changed it to the TCP transport protocol. Having done so, I noticed that ExpressVPN’s speeds dropped to almost exactly the same speeds as VyprVPN.

All speed tests were done using speedtest.net, at what it considered the “optimum server” for the test, based on the VPN endpoint chosen. These tests were done at several different endpoints in both the US and Canada and I used endpoints for both VPNs that were in the same cities. (I’m currently in Eastern Europe.)

My conclusion is that VyprVPN is taking a severe speed hit because it, apparently, defaults to the TCP transport protocol. Similar testing done earlier on using OpenVPN on my ASUS router disclosed similar speeds, leading me to believe that the router’s settings also default to TCP.

Question:
Is it possible to change a setting somewhere, maybe in a config file, to allow use of UDP as the underlying transport protocol? I love just about everything about VyprVPN except for the speed hit, as it prevents my wife from watching her streaming video without choppiness.

Thanks in advance for any help you can provide.

Jim “JR”

The ISP may identify the VPN protocol and then control the dataflow. You could change the protocol to Chameleon to defend DPI(deep protocol identify).

Roberts,

Please forgive my lack of patience if I’m coming across as snarky or sarcastic, it’s not intentional but my patience has been sorely tried.

The fundamental issue isn’t the ISP throttling the dataflow. The problem is VyprVPN throttling the dataflow.

Again, forgive any lack of patience, but I have been struggling with this for months and months and months and months and months. I’ve opened support tickets and aside from some stock suggestions that are totally inapplicable, I have accomplished nothing.

================================

Let me try again and please pay careful attention.

Issue:
When attempting to connect to various VyprVPN endpoints in North America from Moscow Russia, speeds are so slow as to be unusable.

Requirements

  • The main reasons we need a VPN are:
    • Access to our e-mail hosted on Spectrum/Charter servers.
    • Access to subscription streaming video that my wife watches.
  • Access to these items absolutely requires a connection within North America, preferably within the United States. Connections to other endpoints, regardless of location, do not work.
    • Ergo, asking us to try a different endpoint somewhere else may be instructive, but does not help solve the problem.
  • A secondary reason is to avoid any potential political issues while inside Russia.

Setup and test configuration information:

  • All of the information gathered below was gathered using a HP EliteBook 8750p.

    • It is running Windows 7, 64 bit on an i7 3740QM processor with 16 gigs of memory.
    • There is plenty of memory and lots of system resources available for anything you may wish to do.
    • If I can run Microsoft FlightSim-X, or Flight Gear, with decent frame rates, (30 fps+), a simple VPN should be no problem whatsoever. Ergo, system resource limits or processor lag are absolutely not a problem.
  • All tests were run using the speed test facilities provided by Speedtest.net, through both their website, (https://www.speedtest.net), and their downloadable application.

    • All tests allowed Speedtest.net to select the best and fastest server to the endpoint connection.  Note that this did not show any significant variation in speeds.
    • This should provide an accurate test of speeds both within a browser connection, (like my wife’s streaming video), and via a native application, (like our e-mail clients), attempting to use the Internet.
  • All tests were run using the standard available endpoints in New York, Canada, (Toronto if available as a selection), and Washington DC.

  • All tests were run using both VyprVPN and ExpressVPN.

    • Unless otherwise noted, all tests were taken with the “as installed” default configuration for each VPN client.
  • All tests were taken after a complete uninstall of the previous client  If the option for a “complete uninstall” or “remove user data and preferences”, this option was selected.

    • All artifacts from any previous install of a VPN client were removed prior to installation of the next client.
    • This included residual files, (specifically TAP adapter files left behind in %SYSTEM%\system32\drivers), temp files, files in &APPDATA%\local and %APPDATA&\roaming, and any residual files in %PROGRAMDATA%
    • This also included any residual Registry entries left behind after uninstall.
  • I have a fiber-optic Internet connection with Centroset as my ISP in Moscow Russia.

  • I am paying for a 50 mbit symmetrical connection.  (i.e. Upload and download speeds are both 50 mbit.)

    • This is provably so, verified by repeated tests using Speedtest.net.  On average, my line speeds are on the order of 47 mbit with the variation being between 45 mbit and 55 mbit when connected to the specified endpoints in North America, without using a VPN of any kind.

    • These speeds include the NAT traversal of my ASUS RT-N66U router and it’s internal firewall.

    • I have not, nor do I want to, try these tests directly connected to the internet.

  • These observations were taken today, just prior to leaving this message. to ensure accuracy.

  • Observations using ExpressVPN were taken using the latest downloadable version.

  • Observations using VyprVPN were taken using the previous downloaded version, (VyprVPN-2.16.2.8727-installer.exe), as the latest update installed today did not work and would not maintain a stable connection.  (I have opened a separate support ticket for this issue)

================================

Observations while using VyprVPN:

  • Using Chameleon as the VPN protocol:

    • Typical download line speeds vary between 1 and 5 mbit.
    • The median speed, (what I see the most often) is between 1 and 2 mbit, with the majority of the readings closer to 1 mbit than two.  Occasionally speeds drop to less than 1 mbit.
    • Typical upload line speed vary between 56 kbit, (dial-up speeds), and 1 mbit.  Median speeds are in the range of 600 to 850 kbit, but often drop to speeds that are significantly slower.
  • Using OpenVPN (256) as the VPN protocol:

    • Typical download line speeds vary between 10 and 20 mbit.
    • Median speeds are not available because the variation while testing is too great. (i.e. during testing the speeds varied wildly between these two points.
    • Typical upload speeds were in the range of 2 to 5 mbit, with occasional excursions to 20 mbit.
  • The consequence of these speeds and variations is that my wife cannot watch any of her streaming video services without continuous dropouts, slow download stalling, and dropped connections to the video stream.

  • E-mail is unaffected and usable regardless of endpoint selected.  Occasional extremely low speed excursions cause my e-mail to disconnect briefly.

=================================

Observations using ExpressVPN:

  • A chameleon-like protocol is not available.

    • The available choices are “UDP” and “TCP” transport options within the OpenVPN protocol.
  • Using OpenVPN (256) with UDP as the underlying transport protocol:

    • Download speeds were in the range of 40 to 45 mbit with a median speed of about 42-43 mbit.
    • Uploads speeds were in the range of 20 - 30 mbit, with speed excursions to 40+ mbit. The median speed was about 25 mbit.
  • Using OpenVPN (256) with TCP as the underlying transport protocol:

    • Speeds dropped to the equivelent of VyprVPN using OpenVPN, except that the download speed was much more stable.
  • These speeds are sufficiently faster and more stable such that my wife can watch her streaming videos.

  • The big disadvantage with ExppressVPN is that - as of now - it is virtually unusable for e-mail as Charter/Spectrum has blocked virtually all their endpoint IP addresses.  The only usable endpoint for e-mail is Toronto Canada.  They claim to be working on that.  So far, I have not seen any improvement after I mentioned it a week ago.

==================================

Conclusion:

  • Based on my continuing observations with VyprVPN, and my brief observations with ExpressVPN, I am forced to the conclusion that the underlying issue is the VyprVPN portal/process itself.
  • I have not yet had the chance to test with any other VPN provider yet.

I have been a user of VyprVPN since before Golden Frog was created, and I have been a “brand advocate” of the VyprVPN service for as long as its been around.  I have suggested, advised, and, (when possible), sold the VyprVPN service to clients by helping them subscribe.

In point of fact, my blog, (QA TechTips), awarded Giganews/VyprVPN a “Hot Smokin’ Weapon!” award in April of 2013. (Viz.: https://www.qatechtips.com/2013/04/hot-smokin-weapon-award-for-april-2013.html)

When I mentioned my criteria for this particular award, I said:

  • Do they deliver, as promised, 100% NO BULLSHIT?
  • In the extremely unlikely event that “Shit does Happen,” do they promptly step up, take ownership, and make it right, right away?
  • Do they make product quality and customer satisfaction their highest priority?
  • Is this something I can unashamedly recommend to whomever might be interested?
  • Do they make “uncommon valor a common virtue” by going above and beyond the call of duty to meet the customer’s needs?
  • Do they do this at a price that mere mortals like you and I can afford? And despite that, do they make us feel like we’re big-shots like Warren Buffet or Bill Gates?

This has been true for years, both with Giganews and VyprVPN, and it has been my pleasure to recommend them to whomever I could.

While on these fora, I make whatever efforts I can to defend the honor and efforts of everyone at Golden Frog for their tremendous effort and dedication.
(Viz.: The serivce is still facing connetions issue: discussing service through The Great Firewall)

Unfortunately, with the dramatic drop in speeds, I may not be able to continue as a VyprVPN customer.  This would be a crying shame, and I sincerely hope it doesn’t come to this.

Respectfully submitted;

Jim “JR”

Update:

As of about 5:00 pm Moscow Time, OpenVPN speeds have improved and stabilized.

Note that “Moscow Standard Time” in Russia is +3 from GMT/Zulu time and +8 from Eastern Standard Time in the United States.

  • I’m now seeing consistent 20 to 30 mbit download speeds.
  • Upload speeds are in the 5 to 10 mbit range.
    • Note that for the “HD” quality video streams that are now the most common resolution, a 3 to 5 mbit upload connection speed is pretty much the slowest it can be.  Otherwise what would be an excellent download rate will block on queued “ACK” upload responses and slow down to the upload speed, or less.

On the other hand, Chamelion speeds are still too low to be useful.  I’ve seen about a 30% increase in speed, but it is still way to low to support any kind of streaming protocol.

One relevant question though:
Am I correct in assuming that the whole reason for creating the “Chameleon” protocol is because using OpenVPN allows the ISP to inspect packets and throttle your connection?  If so, you’re saving the ISP’s the trouble!  :confused: :astonished: :wink:

Thanks!

Jim  “JR”

Thank you for the details