The CryptoStorm Speed Test Thread

Post by **parityboy** » Mon Jun 08, 2015 4:01 pm

There have been a few users who have documented that they are being capped at ~37.6Mb/s (4.7MB/s) on a 100Mb/s (12.5MB/s) connection, regardless of the node they connect to. In order to build a picture of the performance available to network members, I've created this thread so we can collate experiences.

Guidelines/Rules

We are trying to build an accurate picture of real world throughput, so here are some rules.

1. (Obvious) Minimise the volume of other network traffic on your connection for the duration of the test.
2. DO NOT use speedtest.net. Since we're trying to measure real world throughput, please use a real world application, such as a web browser doing an HTTP download, BitTorrent client, FTP client (FileZilla) or even curl or wget.
3. Where possible, try to use the node which is geographically closest to you. Technically, with fast enough (and few enough) inter-links this should make no difference, but reality is a different story.
4. Again where possible, use the source file (see below) which is geographically closest to you.

Source Files

Linux Mint 17.1 Cinnamon (.torrent is here also)
LinHost Test Files

!	Message from: parityboy
	First link on LinHost page is no longer broken.

Post by **parityboy** » Mon Jun 08, 2015 4:04 pm

@thread

OK, so I'll go first.

Connection: 8Mb/1Mb ADSL
Exit Node: Turing (London)

Code: Select all

wget -O /dev/null http://speedtest.ams01.softlayer.com/downloads/test100.zip

(SoftLayer, Amsterdam)

2015-06-08 12:00:38 (831 KB/s) - ‘/dev/null’ saved [104874307/104874307]

marzametal · Post by **marzametal** » Tue Jun 09, 2015 5:02 am

How connection is made:

ADSL2+ ~24Mb/1Mb <--- too far from exchange so it's like ADSL actually...
Singapore exit node, Singapore download source (apart from torrent... not sure where that originates from...)
Windows Widget v2.22... UDP 443

Download methods:

download via Firefox 29 between 100kb/s and 120kb/s
download via Internet Download Manager between 650kb/s and 850kb/s
download via Tixati between 850kb/s and 1000kb/s

Additional notes worth mentioning:

browser download is higher on other exit nodes, but only reaches 200kb/s - 250kb/s <--- ISP throttling?
download manager app can hit speeds of 1000kb/s to 1100kb/s and stay at that speed, if I connect to other exit nodes
torrent client can max out and stay at 1100kb/s on other exit nodes
I tried connecting via random UDP port (65501); the only difference I noticed was crisper and quicker webpage retrieval and response/reload

NOTE: download manager app and torrent client can thrash speed up to 1300kb/s, but this is detrimental to my internet connection... it causes dropouts and disconnections. So I introduced a throttle limit @ 1100kb/s

turbz · Post by **turbz** » Tue Jun 09, 2015 8:54 pm

Frankfurt seems to have difficulties connecting to peers when using public torrents, private tracker no problem other than speed being quite modest 6-7MB/s.
With Frankfurt Linhost test file 1000MB from netherlands worked quite well at 8.5-9.0MB/s

And the same file with London is steadily at 3.9MB/s for the whole download, the torrent never went past 70KB/s.

Without CS my torrent speeds were 9MB/s and the linhost test file 10MB/s.

I still remain hopeful that there's some solution to this, or at least a good explanation.

!	Message from: parityboy
	Edited throughput units for clarity.

Post by **parityboy** » Tue Jun 09, 2015 9:11 pm

@turbz

Can you clarify that the speeds you quote are MB/s as opposed to Mb/s? It would be good to keep things consistent for ease of readability. Thanks.

EDIT:
Connection: 8Mb/1Mb ADSL
Exit Node: Cantus (Frankfurt)

KTorrent 4.3.1 downloading the Linux Mint 17.1 torrent averaged 790KB/s (6.3Mb/s)

turbz · Post by **turbz** » Tue Jun 09, 2015 9:35 pm

They are MB/s, at least thats how chrome says.

Post by **parityboy** » Tue Jun 09, 2015 10:13 pm

@turbz

Thanks for that. Also, something else came to mind: some of the exit nodes have 100Mb/s network interfaces, rather than 1Gb/s interfaces. The cap you're seeing on the London node may be a result of that (or perhaps something else). I'll have to poke staff to see which nodes have which interfaces.

turbz · Post by **turbz** » Tue Jun 09, 2015 11:58 pm

Thanks a lot for the quick answers and the effort

sysfu · Post by **sysfu** » Wed Jun 10, 2015 7:45 am

Ran a test today from the Kansas City node to linode test file in Dallas.

¡Qué miserables!

Connection: ADSL 22Mb/2Mb

Code: Select all

curl -o /dev/null http://speedtest.dal01.softlayer.com/downloads/test100.zip

Code: Select all

  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100  100M  100  100M    0     0  47070      0  0:37:08  0:37:08 --:--:-- 67877

!	Message from: parityboy
	Edited to improve layout.

Post by **parityboy** » Wed Jun 10, 2015 4:44 pm

@sysfu

Could you edit your post to include the supported bandwidth (8Mb/s, 24Mb/s, 100Mb/s) of your Internet connection? Many thanks.

Post by **Fermi** » Wed Jun 10, 2015 9:07 pm

Code: Select all

curl -o /dev/null http://mirror.nl.leaseweb.net/speedtest/100mb.bin

Leaseweb Amsterdam

Connection: 200Mb/20Mb Cable

without VPN

Code: Select all

  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100 95.3M  100 95.3M    0     0  21.2M      0  0:00:04  0:00:04 --:--:-- 21.2M

Exit Node: Cantus (Frankfurt)

Code: Select all

  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100 95.3M  100 95.3M    0     0  7347k      0  0:00:13  0:00:13 --:--:-- 6977k

Exit Node: Onyx (Paris)

Code: Select all

  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100 95.3M  100 95.3M    0     0   487k      0  0:03:20  0:03:20 --:--:--  582k

Exit Node: Turing (London)

Code: Select all

  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100 95.3M  100 95.3M    0     0  5026k      0  0:00:19  0:00:19 --:--:-- 3873k

/Fermi

sysfu · Post by **sysfu** » Wed Jun 10, 2015 9:45 pm

parityboy wrote:@sysfu

Could you edit your post to include the supported bandwidth (8Mb/s, 24Mb/s, 100Mb/s) of your Internet connection? Many thanks.

Forgive me if I'm being dense but I no longer see an option to edit the previous post even though I am logged into the forum.

The 'raw' speed of the Internet connection is ~22Mps down and ~2Mps up. Currently running running OpenVPN + OpenBSD 5.7 on an Alix device which serves entire home LAN.

I just tried another quick and dirty download test from a Windows workstation (with a youtube vid playing mind you) and the results were much better.

Code: Select all

aria2c http://speedtest.dal01.softlayer.com/downloads/test100.zip

Download Results:
gid   |stat|avg speed  |path/URI
======+====+===========+=======================================================
4735ca|OK  |   341KiB/s|C:/Users/test/Downloads/test100.zip

OpenVPN cpu utilization on the Alix hovered around 45% during the download. The previous test was done on a Mac running Yosemite.

I'll also try tearing down the Cryptostorm OpenVPN tunnel on the router and connect directly from both Windows and Mac systems using the current Narwhal widget and Viscosity respectively, to see what kind of results I get.

Post by **parityboy** » Wed Jun 10, 2015 10:03 pm

@sysfu

Ahh, yes I forgot the edit option becomes unavailable after a certain amount of time. No problem.

thisis4thepeople · Post by **thisis4thepeople** » Sat Jun 13, 2015 7:50 pm

Code: Select all

curl -o /dev/null http://mirror.nl.leaseweb.net/speedtest/100mb.bin

Exit node: Laika

Code: Select all

  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100 95.3M  100 95.3M    0     0  6208k      0  0:00:15  0:00:15 --:--:-- 7315k

Managed to get around 70 Mbps via a public torrent and 50 on YouTube.

sysfu · Post by **sysfu** » Sat Jun 13, 2015 11:13 pm

Direct via ISP

Code: Select all

aria2c http://speedtest.sea01.softlayer.com/downloads/test100.zip

Download Results:
gid   |stat|avg speed  |path/URI
======+====+===========+=======================================================
ed2189|OK  |   2.6MiB/s|D:/Users/test/Downloads/test100.zip

OpenBSD/Alix router via Seattle cluster:

Code: Select all

Download Results:
gid   |stat|avg speed  |path/URI
======+====+===========+=======================================================
a40207|OK  |   415KiB/s|D:/Users/test/Downloads/test100.zip

Narwhal 2.22 widget via Seattle cluster:

Code: Select all

Download Results:
gid   |stat|avg speed  |path/URI
======+====+===========+=======================================================
463ae2|OK  |   1.1MiB/s|D:/Users/test/Downloads/test100.zip

Post by **parityboy** » Sun Jun 14, 2015 2:13 am

thisis4thepeople wrote:
Code: Select all
curl -o /dev/null http://mirror.nl.leaseweb.net/speedtest/100mb.bin
Exit node: Laika

<snip>

What's your "naked" speed?

sysfu · Post by **sysfu** » Sun Jun 14, 2015 9:33 am

Raw or nekkid connection speed: ADSL ~22Mbps/2Mbps

thisis4thepeople · Post by **thisis4thepeople** » Sun Jun 14, 2015 9:45 am

parityboy wrote:
thisis4thepeople wrote:
Code: Select all
curl -o /dev/null http://mirror.nl.leaseweb.net/speedtest/100mb.bin
Exit node: Laika

<snip>
What's your "naked" speed?

100/100 Mbps.

sysfu · Post by **sysfu** » Tue Jun 16, 2015 9:51 pm

Thought I'd share a pretty cool command line client for speedtest.net that I found the other day: https://github.com/zpeters/speedtest

thisis4thepeople · Post by **thisis4thepeople** » Thu Jun 18, 2015 10:24 am

Another test: One only on YouTube, the other two simultaneous downloads (YouTube + savefrom.net).

67 Mbps on YouTube = 8,375 MB/s
6,8 MB + 16 Mbps = 8,8 MB/s

bricky149 · Post by **bricky149** » Thu Jun 18, 2015 2:33 pm

I have a 152Mbps connection at home although it peaks short of 120Mbps IRL. While I'm connected via CStorm I see a top speed of around 50Mbps. The average speed for torrents is around 2.4MiB/s (unchanged from before my connection upgrade) and for regular files I've seen it peak at around 4MiB/s. I've been messaging through Twitter since it's seen as an edge case, difficult to tune my OS since Windows makes it hard to.

bricky149 · Post by **bricky149** » Fri Jul 03, 2015 1:22 pm

(Can't edit my posts since the forum upgrade so I'll make a new one.)

I've had to switch to a wired connection, with a 20ft Ethernet cable, because my wireless adapters either drop after 5 minutes or cause Windows to malfunction. On torrents I see ~5MiB/s for download and for regular files it's ~6MiB/s. The highest speed I can achieve on a naked connection is ~14MiB/s.

Thracky · Post by **Thracky** » Wed Jul 08, 2015 6:15 am

So I'm in my first few days of cryptostorm and it's pretty painful for me so far. Here's why.

Connection Details:
250 Mb down/20 Mb up - Eastern-ish Canada

All tests performed using wget of http://speedtest.dal01.softlayer.com/do ... est100.zip

No VPN

Code: Select all

100%[===============================================>] 104,874,307 21.4MB/s   in 8.4s   

2015-07-07 20:27:34 (11.9 MB/s) - ‘/dev/null’ saved [104874307/104874307]

File wasn't completely large enough to show my normal sustained speed but the peak at the end is pretty typical and even a bit on the low side.

Montreal Exit - 198.27.76.1

Code: Select all

2015-07-07 20:53:34 (77.0 KB/s) - ‘/dev/null’ saved [104874307/104874307]

So that was pretty awful. Also, I couldn't even keep a persistent SSH or IRC session without constant disconnects. So, let's try a completely different exit...

US Central - 142.54.172.51
Seriously, this was going between 5-20KB/sec if at all so I cancelled it.

NEXT. Let's see if going across the ocean helps...

Iceland - 109.71.42.163

Code: Select all

2015-07-07 21:03:35 (1.93 MB/s) - ‘/dev/null’ saved [104874307/104874307]

Well that was substantially better and it finished strong at just over 4 MB/s. Let's try one more.

London - 130.180.201.117

Code: Select all

2015-07-07 21:01:30 (4.73 MB/s) - ‘/dev/null’ saved [104874307/104874307]

That's more like it.

For the sake of comparison, let's check a couple exit nodes on PIA:
PIA - Toronto (Closest geographically)

Code: Select all

2015-07-07 21:05:43 (3.32 MB/s) - ‘/dev/null’ saved [104874307/104874307]

PIA - US East

Code: Select all

2015-07-07 21:06:54 (2.95 MB/s) - ‘/dev/null’ saved [104874307/104874307]

And one overseas for good measure, let's try London.

PIA - London

Code: Select all

2015-07-07 21:10:05 (1.74 MB/s) - ‘/dev/null’ saved [104874307/104874307]

And that's that.

So, I'm not particularly thrilled with the North American exits that I tried, but at least Iceland and London performed alright.

marzametal · Post by **marzametal** » Wed Jul 08, 2015 8:06 am

Thracky wrote: US Central - 142.54.172.51
Seriously, this was going between 5-20KB/sec if at all so I cancelled it.

This one was a pain for me this morning... the others seem to serve me well.

Post by **parityboy** » Thu Jul 09, 2015 12:16 am

@Thracky

On your faster speed runs via CS, what was your CPU load like?

Thracky · Post by **Thracky** » Thu Jul 09, 2015 2:33 am

parityboy wrote:@Thracky

On your faster speed runs via CS, what was your CPU load like?

One core bouncing around between 9-25% on the openvpn process. (i7-2600k, no overclock)

Post by **parityboy** » Thu Jul 09, 2015 4:18 am

Thracky wrote:
parityboy wrote:@Thracky

On your faster speed runs via CS, what was your CPU load like?
One core bouncing around between 9-25% on the openvpn process. (i7-2600k, no overclock)

Well, 25% load tells me the issue certainly isn't at your end. I'm beginning to wonder if the issue is the OpenVPN process itself, or maybe further down the networking stack? Considering the fact that Linux-based routers are capable of throughput far greater than 50Mbit/s - combined with the fact that CS' exit nodes are full-blown x86 servers - I'd say the throughput issue lies with OpenVPN itself

However, without proper testing I'm just guessing.

sysfu · Post by **sysfu** » Sun Aug 02, 2015 11:03 pm

Using exit linux-cantus1.xn--cdaan2d.be46.165.222.248
Native 22Mbps/2Mbps dn/up

Code: Select all

Retrieving speedtest.net configuration...
Retrieving speedtest.net server list...
Testing from Leaseweb Germany GmbH (46.165.222.248)...
Selecting best server based on latency...
Hosted by Base-Mail (Frankfurt) [100.73 km]: 237.964 ms
Testing download speed........................................
Download: 8.62 Mbit/s
Testing upload speed..................................................
Upload: 0.98 Mbit/s

Post by **parityboy** » Tue Aug 04, 2015 4:59 am

@sysfu

From the OP

2. DO NOT use speedtest.net. Since we're trying to measure real world throughput, please use a real world application, such as a web browser doing an HTTP download, BitTorrent client, FTP client (FileZilla) or even curl or wget.

thisis4thepeople · Post by **thisis4thepeople** » Tue Oct 06, 2015 1:38 pm

64 Mbit/s on UK server on a 100 Mbit/s line.

DonderPjem · Post by **DonderPjem** » Wed Oct 07, 2015 8:28 am

My tradeoff for using CS is almost 90% of my bandwidth. My usually 100mbit up/down is reduced to 10~12mbit. Fortunately I only have to endure this during internet browsing as I outsource my heavy torrenting to a seedbox. But it still becomes noticable when trying to watch yt videos in 1440p or download something larger than a few megs.

It looks like I am being hit the hardest here, does anyone have optimalisation tips? Should I just hope for a server cluster in my country?

thisis4thepeople · Post by **thisis4thepeople** » Wed Oct 07, 2015 11:02 am

Preferably, you'd choose a server close to you. Test all servers closest to you and use the one that gives you the highest speed, if that is what you prefer.

DonderPjem · Post by **DonderPjem** » Wed Oct 07, 2015 5:44 pm

Im in the Netherlands and the london, frankfurt and paris clusters all give the same performance.

Post by **parityboy** » Wed Oct 07, 2015 10:39 pm

@DonderPjem

Can you give a little more details as to your setup?

1. Is the VPN endpoint on your router or on your desktop? What's the respective CPU load?
2. What CPU/OS are you running?

Thanks.

DonderPjem · Post by **DonderPjem** » Fri Oct 09, 2015 9:57 pm

I'm using the cryptostorm widget on my 64bit windows machine with an i7 2630QM. The cpu usage is insignificant. I'm pretty sure the bottleneck is not on my part. PIA gave me speeds at around 95% of my native bandwidth. Ive tried every server cluster by now and I havent been able to utilize more than 15% of my 100mbit connection yet.

Post by **parityboy** » Sat Oct 10, 2015 4:44 pm

@DonderPjem

Thanks for the info.

Would it be possible for you to conduct a couple of tests using the links at the top of the page? Basically, connect to an exit node closest to your location, then pick a file source for Linux Mint which is also closest to your location - so for example if you are in Germany, you would use the Frankfurt exit node and download the ISO from this link.

If you could report back with the results of a couple of runs, it would be appreciated.

DonderPjem · Post by **DonderPjem** » Sun Oct 11, 2015 12:45 am

Thanks for your time!
Ive tried every mirror with the respective exit node and note down the maximum speed it maintained for about 10 seconds in the first 2 minutes.

(speeds are in KB/s and mirrors are in order of appearance on http://www.linuxmint.com/edition.php?id=172)

france
300
250
400
250
600

germany
300
900
200
350
300
700
600

uk
400
offline

switzerland
550

netherlands (native connection)
3000
2200

netherlands (germany exit node)
500
350

JTD121 · Post by **JTD121** » Mon Oct 12, 2015 9:16 pm

I definitely want to help out; couple questions. Why do we avoid speedtest.net? Probably because it's a bursty test, rather than an average or median test?

Are there any particular software to track this on the user end? I know uTorrent has a sort of speed tracker over the download portion of a torrent, but outside of torrents?

Currently using the eyeball method while watching an HTTP download from a mirror.

JTD121 · Post by **JTD121** » Tue Oct 13, 2015 12:49 am

To borrow/adapt marzametal's format:

How connection is made:

Comcast cable; rated for 25/5, though real-world it's 30/6 on-the-wire (isn't Comcast generous?)
Asus RT-N66U running Advanced Tomato 2.9-K131
Wireless desktop, using 5GHz 802.11N, reads full 450Mbps connection
Test 1: Kansas City - US Central node
Test 2: Toyko - Japan (because Japan )
Windows Widget v2.22: I believe random UDP port; currently 41217. All 3 in Security tab checked
Single download at a time (one HTTP download, one torrent, etc)

Download methods: Linux Mint 17.2 Cinnamon 64-bit (using KB = kilobyte, kb= kilobit as reported by software)

Test 1: From University of Oklahoma mirror
Also streaming from Digitally Imported at 128k AAC during tests (so ~130-150 kb/s with overhead); loves me some EDM
Download via Firefox 41.0.1 between 31.2KB/s (absolute minimum observed) and 499KB/s (absolute maximum observed); also notice a dip in speed when loading up this thread, or other small (non-heavy interactive/download) sites (Slashdot (though not too light anymore), Twitter, Gmail (Ctrl+F5 refresh)
Speeds seem to hover in the 100-250KB/sec range, jumping up and down every so often.

Test 2: Same mirror (Univ of OK)
Speeds as low as 5.5KB/sec, almost never getting higher than 60KB/sec
Talk about stress-test, huh?
Same download setup as before, just different node.
Made DI.FM stream drop out repeatedly, while also affecting the download speed of the ISO

Disconnecting from CS, I see that same download, from the same mirror, it's superluminal; pegging the connection at 2.5-3.6MB/sec (hovering at 2.8-3.4MB/sec). This information is to help the network, so please don't take it as a detrimental post I've created here. Not sure the exact setup for connectivity at the node, within your kernels, etc. Just for point of reference.

Will update with additional information as I run tests (torrents, different nodes, different mirrors)

rwolslegel · Post by **rwolslegel** » Tue Oct 13, 2015 2:04 am

I spent lots of time running tests and have generated a ton of data. Here it is:

All tests were done within about 30 minutes of each other using wget with five different servers and three different hosting providers. All were done from GNU/Linux. I am physically located on the eastern side of the United States.

Softlayer - Dallas US:

Not Connected to CS: 29.2MB/s
US Central: 3.03MB/s
US West: 1.2MB/s
Montreal: 6.90MB/s
London: 7.23MB/s
Switzerland: 1.04MB/s

Softlayer - Seattle US:

Not Connected to CS: 20.3MB/s
US Central: 2.17MB/s
US West: 2.11MB/s
Montreal: 7.64MB/s
London: 5.85MB/s
Switzerland: 1019KB/s

Softlayer - Amsterdam NL:

Not Connected to CS: 19.0MB/s
US Central: 2.52MB/s
US West: 632KB/s
Montreal: 4.6MB/s
London: 10.8MB/s
Switzerland: 1.2MB/s

Linode - Atlanta US:

Not Connected to CS: 24.3MB/s
US Central: 1.78MB/s
US West: 1.74MB/s
Montreal: 5.47MB/s
London: 7.01MB/s
Switzerland: 434KB/s

DigitalOcean - New York US (A VPS that I spun up just to experiment):

Not Connected to CS: 28.3MB/s
US Central: 1.73MB/s
US West: 1.67MB/s
Montreal: 5.84MB/s
London: 5.97MB/s
Switzerland: 230KB/s

Some conclusions that I can draw from this is that the fastest node for me is either Montreal or London, while the node that's actually closest to me is one of the slowest (uscentral). This is consistent with other observations I have made in the past, for example being unable to stream 1080p Youtube when connected to US nodes.

Post by **parityboy** » Tue Oct 13, 2015 4:00 am

@rwolslegel,@JTD121,@DonderPjem

Many thanks for posting your experiences in this thread. It would appear that routing across the United States is awful, at least for the routes used by the datacentres where Cryptostorm nodes are located. The paths to Switzerland aren't too hot either, by all accounts.

Re: speedtest.net, yes it's bursty, also browsers running a speed test written in Flash(!) aren't the most reliable things in the world. File transfer software such as FileZilla or wget (for the 'nix heads out there) gives far more accurate results.

JTD121 · Post by **JTD121** » Sun Oct 18, 2015 5:22 pm

Can't seem to edit my post above, but this is Test 3, connected to London - England node.

Test 3: Univ of Kent UK (geographically close to London node (I assume))
Speeds don't seem to drop as low as previous US-central node
Speeds from 75KB/sec to 1.5MB/sec; mostly seem to fluctuate in the 150 ~ 450KB/sec range
Prolonged idle (reading a single page of something) seems to give download a bit of a boost; observed increases between 600KB/s to 750KB/sec, several bursts 30~40 second to 1.3 - 1.5MB/sec
Same download setup as before, just different node.
DI.FM stream didn't skip a beat (har har!) during download

JTD121 · Post by **JTD121** » Thu Nov 19, 2015 10:30 pm

So with the new nodes and replacement nodes up and running, and using gigabit pipes, do we want to retest some of them?

Post by **parityboy** » Thu Nov 19, 2015 11:34 pm

@JTD121

Might be an idea. My connection is too slow to make any difference, but for those with 30Mb+ connections, it might prove useful.

EDIT:
Taken from this post:

Code: Select all

france:                Speed: 1000Mb/s
frankfurt:             Speed: 100Mb/s
london:                Speed: 1000Mb/s
france#2:              Speed: 1000Mb/s
lisbon:                Speed: 1000Mb/s
seattle:               Speed: 1000Mb/s
singapore:             Speed: 100Mb/s
switzerland:           Speed: 1000Mb/s
italy:                 Speed: 100Mb/s
vegas:                 Speed: 1000Mb/s
canada:                Speed: 100Mb/s
moldova:               Speed: 100Mb/s

JTD121 · Post by **JTD121** » Sat Nov 21, 2015 2:24 am

Hm...My connection is rated about 25/5, and I've regularly seen speeds around 31/7, so I'll give this a couple test downloads soon.

I'll try to keep similar format as before (again borrowing marzametal's format).

JTD121 · Post by **JTD121** » Sat Nov 21, 2015 6:27 pm

This test site was suggested in IRC, and it looks like it's not Flash-based. Also, has an adjustable long-term speed testing method. Currently testing it out without CS to see how it fares.

Will obviously do some stress testing and compare to CS as well.

JTD121 · Post by **JTD121** » Wed Nov 25, 2015 9:20 am

Still can't seem to edit my posts above, but this is Test 4, connected to Buffalo, NY node node.

Test 4: GOG Downloader; downloading ~830MB in total
Speeds don't seem to fluctuate as in previous tests
Speeds from 650KB/sec to 1.4-1.5MB/sec; Very consistent staying towards the top of this range
Prolonged idle (reading a single page of something) doesn't seem to affect overall bandwidth.
Also running a multi-threaded test at http://testmy.net for extra bandwidth loading
Not really using all bandwidth available to me without CS VPN running, but definitely getting better!
No DI.FM streaming this time

vehf277 · Post by **vehf277** » Wed Dec 02, 2015 9:42 pm

What about testmy.net and speedof.me for speed test usage?

Post by **parityboy** » Fri Jan 01, 2016 10:31 pm

@thread

Internet Connection: Cable, 50Mb/s
Local Connection: 802.11n, 72Mb/s as tested

Code: Select all

Node           wget       Torrent (avg)    Torrent (peak)
Naked          5.10MB/s   4.35MiB/s        4.99MiB/s
Cantus         3.74MB/s   3.76MiB/s        5.10MiB/s
Jord           2.60MB/s   3.57MiB/s        4.43MiB/s
Tagus          2.63MB/s   3.15MiB/s        3.48MiB/s
Turing         4.58MB/s   3.88MiB/s        4.53MiB/s

speedtesting · Post by **speedtesting** » Sat Jan 16, 2016 5:29 pm

Bare connection = 40Mbp/s by provider's router in modem only mode.
Local = ethernet cable - netgear switch - 500Mbp/s powerline - ASUS RT-AC68R/U (Shibby Tomato 132 running OpenVPN. Client 2 being used for this test).

Tested from UK.

----------

London.

$ wget -O /dev/null http://releases.ubuntu.com/14.04.3/ubun ... -amd64.iso
--2016-01-16 11:44:00

100%[============================================================================================================================================================>] 601,882,624 2.67MB/s in 3m 20s

2016-01-16 11:47:20 (2.87 MB/s) - ‘/dev/null’ saved [601882624/601882624]

-----------

Paris.

$ wget -O /dev/null http://releases.ubuntu.com/14.04.3/ubun ... -amd64.iso
--2016-01-16 11:48:36--

100%[============================================================================================================================================================>] 601,882,624 2.79MB/s in 4m 55s

2016-01-16 11:53:32 (1.94 MB/s) - ‘/dev/null’ saved [601882624/601882624]

----------

Frankfurt.

$ wget -O /dev/null http://releases.ubuntu.com/14.04.3/ubun ... -amd64.iso
--2016-01-16 11:56:12--

100%[============================================================================================================================================================>] 601,882,624 2.22MB/s in 4m 38s

2016-01-16 12:00:50 (2.06 MB/s) - ‘/dev/null’ saved [601882624/601882624]

-----------

Switzerland.

*$ wget -O /dev/null http://releases.ubuntu.com/14.04.3/ubun ... -amd64.iso
--2016-01-16 12:02:47--

100%[============================================================================================================================================================>] 601,882,624 2.95MB/s in 5m 39s

2016-01-16 12:08:26 (1.69 MB/s) - ‘/dev/null’ saved [601882624/601882624]

cumpsd · Post by **cumpsd** » Sun Apr 03, 2016 1:30 am

Strange results...

Testing by downloading http://speedtest.ams01.softlayer.com/do ... est100.zip via wget

Without VPN / Belgium:
240Mbps down / 20 Mbps up
2016-04-02 00:56:00 (25.6 MB/s) - 'test100.zip' saved [104874307/104874307]

Paris UDP:
2016-04-02 01:03:00 (259 KB/s) - 'test100.zip.1' saved [104874307/104874307]
Paris TCP:
2016-04-02 02:00:22 (2.06 MB/s) - 'test100.zip.11' saved [104874307/104874307]

Frankfurt UDP:
2016-04-02 01:18:18 (269 KB/s) - 'test100.zip.3' saved [104874307/104874307]
Frankfurt TCP:
2016-04-02 01:58:56 (1.37 MB/s) - 'test100.zip.10' saved [104874307/104874307]

Switzerland UDP:
2016-04-02 01:32:14 (267 KB/s) - 'test100.zip.5' saved [104874307/104874307]
Switzerland TCP:
2016-04-02 02:02:35 (7.21 MB/s) - 'test100.zip.12' saved [104874307/104874307]

Lisbon UDP:
2016-04-02 01:11:07 (233 KB/s) - 'test100.zip.2' saved [104874307/104874307]

Rome UDP:
2016-04-02 01:25:20 (277 KB/s) - 'test100.zip.4' saved [104874307/104874307]

Voodoo-USWest-Iceland:
2016-04-02 01:46:17 (287 KB/s) - Read error at byte 103115450/104874307 (Invalid argument). Retrying.

Voodoo-USSouth-IsleOfMan:
No results, kept reconnecting and was very unstable

thisis4thepeople · Post by **thisis4thepeople** » Fri May 06, 2016 5:50 pm

Managed a new record today, I wasn't even paying attention when I heard the fans of my device go on and checked to see what was going on. To my surprise, I was uploading at quite a nice speed, around 75 Mbps.

Connected to the Oslo node by the way.

thisis4thepeople · Post by **thisis4thepeople** » Sun Jun 19, 2016 6:16 pm

New day, new record (Switzerland server): 13.26 MB/s equates to over 100 Mbit/s (106 to be precise). This is on a 200 Mbit/s line if I'm not mistaken. My ISP gave me 100 Mbit/s before but I think it has been upgraded in the meantime.

Post by **parityboy** » Wed Jun 29, 2016 5:23 am

@thisis4thepeople

Many thanks for keeping us updated with your results, it's greatly appreciated. Unfortunately, I'm still stuck on an 8/1 ADSL line, so I'm unable to contribute further.

turbz · Post by **turbz** » Tue Jul 12, 2016 12:06 am

Well, finally I get acceptable speeds

Constant +18MBps in and +2MBps out, this is on 350/30 line.
Switzerland node.

Post by **parityboy** » Tue Jul 12, 2016 3:11 am

@turbz

Many thanks for your post, it's nice to see people still contributing to this thread. Proves that secure VPNs don't have to be slow.

thisis4thepeople · Post by **thisis4thepeople** » Fri Sep 02, 2016 2:27 pm

I'm back with new results. Last time I had a down-stream of over 13 MB/s, now it's the other way around:

Not to shabby. This was via Denmark by the way.

Post by **parityboy** » Mon Sep 05, 2016 12:02 am

@thisis4thepeople

Many thanks for keeping us up-to-date with throughput results, it's greatly appreciated.

If only others would chime in...

sm1th · Post by **sm1th** » Wed Sep 07, 2016 2:04 pm

Ok, I'll bite.

40Mb bareback connection, over 5ghz wifi via Asus RT-AC68U (overclocked to 1200/800). England node. I've used the speedtest file from a post above...for some vague semblance of comparison (or I'm just lazy).

Code: Select all

$ wget -O /dev/null http://speedtest.ams01.softlayer.com/downloads/test100.zip

100%[======================================================================================================================================>] 104,874,307 2.92MB/s   in 36s    

2016-09-07 09:50:29 (2.77 MB/s) - ‘/dev/null’ saved [104874307/104874307]

Post by **parityboy** » Sun Sep 11, 2016 5:07 pm

@sm1th

Many thanks for sharing this. Your numbers look in line with what you connection can support.

wpaschukat · Post by **wpaschukat** » Sun Sep 11, 2016 6:03 pm

Might as well..

Paris node, on a 100/10 line:

Code: Select all

wget -O /dev/null http://speedtest.ams01.softlayer.com/downloads/test100.zip
HTTP request sent, awaiting response... 200 OK
Length: 104874307 (100M) [application/zip]
Saving to: ‘/dev/null’

/dev/null                                    100%[==============================================================================================>] 100.02M  6.42MB/s   in 17s

More than reasonable

!	Message from: parityboy
	Edited to improve formatting

Khariz · Post by **Khariz** » Tue Sep 13, 2016 11:03 am

My speeds are great. I just tested all the U.S. Servers and England and I'm getting 45down, 35up on my 50/50 connection.

I'm even getting 30down on the Isle of Man voodoo.

thisis4thepeople · Post by **thisis4thepeople** » Thu Oct 06, 2016 5:48 pm

The Switzerland server is still speedy:

Post by **parityboy** » Fri Oct 07, 2016 1:39 am

@thisis4thepeople

I was tempted to rent a 1Gb/s dedi and install a desktop on it just to get some numbers for my own tests, but I see there's no need. CS nodes are certainly fast enough to support the user base.

thisis4thepeople · Post by **thisis4thepeople** » Sun Nov 20, 2016 1:11 pm

Switzerland server, this time from the command line with speedtest-cli.

Code: Select all

Retrieving speedtest.net configuration...
Testing from DataSource (XXX.XX.XXX.XX)...
Retrieving speedtest.net server list...
Selecting best server based on ping...
Hosted by Datasource. AG (Zug) [6.26 km]: 51.064 ms
Testing download speed................................................................................
Download: 78.53 Mbit/s
Testing upload speed....................................................................................................
Upload: 39.13 Mbit/s

tenaciousj · Post by **tenaciousj** » Tue Nov 29, 2016 12:18 pm

I ran several tests from Charlotte, NC and got pretty good results on a
300Mbps cable connection routed over 802.11ac to the client
I normally get 150-200Mbps wifi speeds, so pulling 100Mbps on speedtest over VPN is pretty damn fast

large file wget download results:
(connected to the std. US East connection)

100mb file:
atlanta 3.71MB/s time 30s
newark 7.41MB/s time 13s
dallas 4.74MB/s time 21s

1000mb from Leaseweb in Virginia gave me 5.26MB/s time: 3m 1s

speedtest.net (for kicks):
100Mbps down 16.42Mbps up (up is capped at 20)

cryptocornhole · Post by **cryptocornhole** » Thu Feb 02, 2017 11:02 pm

Here are my results:

Code: Select all

wget -O /dev/null http://speedtest.ams01.softlayer.com/downloads/test100.zip

No VPN:

Code: Select all

2017-02-02 10:45:25 (3.65 MB/s) - ‘/dev/null’ saved [104874307/104874307]

cstorm_USsouth:

Code: Select all

2017-02-02 10:44:39 (1.07 MB/s) - ‘/dev/null’ saved [104874307/104874307]

cstorm_USnorth:

Code: Select all

2017-02-02 10:50:38 (1.67 MB/s) - ‘/dev/null’ saved [104874307/104874307]

cstorm_USeast:

Code: Select all

2017-02-02 10:52:59 (1.85 MB/s) - ‘/dev/null’ saved [104874307/104874307]

cstorm_UScentral:

Code: Select all

2017-02-02 10:55:52 (812 KB/s) - ‘/dev/null’ saved [104874307/104874307]

Russell · Post by **Russell** » Wed Mar 22, 2017 12:33 am

I guess it's my turn now.

How connection is being made: Duplex FTTB 25Mb/25Mb cable with cheap TP-LINK router (very stable commection. Ping between three local cities is only 9ms).
Exit nodes: Using Swiss and German exit nodes for testing.
OS: Windows 7.

Speedtest (ping/dl speed/upload speed): Original - 3ms/11.85Mbs/11.33Mbs (it's evening time, switch is probably overloaded)
Swiss node (My provider isn't hyper-restrictive, so I mainly use UDP) - 71ms/4.50Mbs/7.39Mbs (pretty decent result considering the fact that Swiss isn't that close to me)
German node (Same story - UDP, Frankfurt) - 92ms/2.92Mbs/8.87Mbs (it's LeaseWeb, so no wonder)
German node (UDP, Dusseldorf) - 97/3.8Mbs/5.04Mbs (Nothing unusual, typical German node).

I only use these as my daily driver because they're located FAIRLY close to me.

Yesterday I've had 52ms for ping and 10Mb/s for both dl/upload at Swiss node.

I usually get 1.4MB/s during torrent download. That's more than enough for me. If I'll need high speed for multimedia, I have a dedicated line and router for that.

!	Message from: parityboy
	Edit throughput units for clarity.

JTD121 · Post by **JTD121** » Thu Mar 30, 2017 4:59 pm

Just messin' around, and used DSLReports speed test! 60s test time, 32 download/upload streams (not sure how this affects the test).

I believe this was connected to the US North or US East node cluster....

My speed tier is 200mbps down, 10mbps up. Using an Arris SB6190 with the Puma6 issues.

Post by **parityboy** » Sat Apr 08, 2017 6:58 pm

@JTD121

Nice to see that the upstream throughput is inline with your "naked" package performance. Assuming the VPN connection starts on your PC, the downstream throughput could possibly point to congestion on that node.

Is that node the closest one to your location?

frankfooter · Post by **frankfooter** » Sun May 07, 2017 6:23 am

25mbit DSL
US North Node
Downloading http://cachefly.cachefly.net/100mb.test
900KB/sec

Post by **parityboy** » Fri Jun 02, 2017 8:23 am

Connection: 4G
Exit Node: England

Code: Select all

wget http://cachefly.cachefly.net/100mb.test
--2017-06-02 04:19:36--  http://cachefly.cachefly.net/100mb.test
Resolving cachefly.cachefly.net (cachefly.cachefly.net)... 205.234.175.175
Connecting to cachefly.cachefly.net (cachefly.cachefly.net)|205.234.175.175|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 104857600 (100M) [application/octet-stream]
Saving to: ‘100mb.test.2’

100%[================================================================================================>] 104,857,600 3.36MB/s   in 31s    

2017-06-02 04:20:12 (3.25 MB/s) - ‘100mb.test.2’ saved [104857600/104857600]

blurb · Post by **blurb** » Wed Oct 17, 2018 3:13 pm

Post upgrade 'ECC':

VPN connected to ASUS RT-AC68U ('FreshTomato' firmware) overclocked to 1200/800.
OpenVPN 2.4.6, OpenSSL 1.0.2p
Base connection is ~40 Mbits down, 10 Mbits up. VDSL.
Connected to closest exit node.

Test was conducted on an old (and much loved) thinkpad connected to router via ethernet. I'm not bothering with a wifi test, too many variables for it to mean much imho.

$ ./speedtest-cli
Retrieving speedtest.net configuration...
Testing from UK Dedicated Servers Limited (5.101.149.4)...
Retrieving speedtest.net server list...
Selecting best server based on ping...
Hosted by Uno (Milton Keynes) [49.22 km]: 14.871 ms
Testing download speed................................................................................
Download: 33.77 Mbit/s
Testing upload speed................................................................................................
Upload: 8.95 Mbit/s

$ ./speedtest-cli
Retrieving speedtest.net configuration...
Testing from UK Dedicated Servers Limited (5.101.149.4)...
Retrieving speedtest.net server list...
Selecting best server based on ping...
Hosted by Uno (Milton Keynes) [49.22 km]: 15.385 ms
Testing download speed................................................................................
Download: 34.43 Mbit/s
Testing upload speed................................................................................................
Upload: 8.93 Mbit/s

$ ./speedtest-cli
Retrieving speedtest.net configuration...
Testing from UK Dedicated Servers Limited (5.101.149.4)...
Retrieving speedtest.net server list...
Selecting best server based on ping...
Hosted by Uno (Milton Keynes) [49.22 km]: 15.958 ms
Testing download speed................................................................................
Download: 34.74 Mbit/s
Testing upload speed................................................................................................
Upload: 8.86 Mbit/s

I thought the new fancy ways would be a bit slower, but it seems not.

Post by **parityboy** » Wed Oct 17, 2018 6:48 pm

@blurb

Many thanks for sharing this.

Ironically, while the newer ECC cryptographic algorithms are much more efficient than the older RSA - requiring much less CPU power for the same level of cryptographic strength - there's a good chance that the older routers (which would actually benefit from this) will never be updated to take advantage of it.

blurb · Post by **blurb** » Tue Oct 30, 2018 2:11 pm

Question: Is a cheap vps set up using https://github.com/StreisandEffect/streisand faster than Cryptostorm? To make it a bit unfair, I'm connecting directly from my laptop and not the router.

Comparison is to my last post in this thread.

$ ./speedtest-cli
Retrieving speedtest.net configuration...
Testing from Linode (***.***.***.***)...
Retrieving speedtest.net server list...
Selecting best server based on ping...
Hosted by 23media GmbH (Frankfurt) [0.53 km]: 31.704 ms
Testing download speed.........................................................
Download: 34.58 Mbit/s
Testing upload speed...........................................................
Upload: 8.87 Mbit/s

Answer: Not really, no, not on my old laptop.

Just thought I'd share.