Speed comparison IPv4 <-> IPv6

[roberttk]

Hello,

I have a 100/5M cable internet connection and I decided to test a HE tunnel.
The nearest tunnel server is in Frankfurt.
Ping:

Code Select Expand

rtt min/avg/max/mdev = 13.099/16.674/27.818/3.647 ms

I've also tried a tunnel server in Amsterdam:

Code Select Expand

rtt min/avg/max/mdev = 12.850/17.350/29.934/4.040 ms

Both with the exactly same result:
I only get about 40 mbps using IPv6.






I've also done multiple test downloads (500 MB files) from my servers in Germany, France and the Netherlands with the exact same result - I'm capped at about 40 mbps.
Example:

Code Select Expand

IPv4:
2015-03-23 21:46:34 (12,0 MB/s) - `500MB.test' saved [500000000/500000000]

Code Select Expand

IPv6:
2015-03-23 21:49:37 (4,61 MB/s) - `500MB.test' saved [500000000/500000000]

Is there anything I can configure to increase the speed?
I am able to get the full speed on HE connected data centers, so it's not the connection from Unitymedia to HE.

If someone else has a 100 mbps home connection, it would be great to see a comparison done on ipv6-test.com.

Thank you!

[JRMTL]

I find ipv6-test.com speed tests fluctuate wildly even using the same server. I can definitely confirm speeds > 100 mbps. I regularly transfer data at 100+ mbps. My line rate is 175/175 and I always have speed tests ~ 174/178 with the exception of when my ISP profile was off and I was able to pull in 239/176.

[evantkh]

I have no idea on the total bandwidth of the tunnel servers, but I can view YouTube HD 1080p videos correctly with using the tunnel server in Hong Kong(the non-BGP one).

I think the speed test results are not important. The actual performance in using the internet is more important because websites usually transfer small files but speedtest results are based on transferring large files.

Same reason, the 3G HSPA+ performance of my phone is similar to that of others' 4G(my phone is a 3G phone).

[roberttk]

I find ipv6-test.com speed tests fluctuate wildly even using the same server.

That's true, but it doesn't matter, because I can confirm I'm limited at about 40 mbps and even while downloading multiple files from multiple servers at the same time, I am not able to get more than 40M.

I think the speed test results are not important. The actual performance in using the internet is more important because websites usually transfer small files but speedtest results are based on transferring large files.

Well, I don't agree.
When downloading bigger files (500MB+) over http I'm now capped at 40 mbps rather than 100 mbps, which means it takes about 2.5 the time to download the same file.


Does my provider throttle the tunnel connection?
It would be great to see test results with a similar async internet connection (like 100/5 or 100/10).

[evantkh]

Perhaps you can try to contact ipv6@he.net. However, HE does not provide a specification on bandwidth of each tunnel.

I use the non-BGP tunnel server in Hong Kong, I tested the maximum speed is about 50M and the average is enough for streaming HD videos, although not 4k.

[roberttk]

It doesn't matter which tunnel I use.
Here are some more results (the Comcast test does not work unfortunately).

IPv4 UK:

IPv6 UK:

Even with 6 simultaneous downloads I'm not getting more than 47.6 Mbps.

Another comparison (Amsterdam based server):


Test from my provider:

[broquea]

The tserv is shared resources, so you compete for bandwidth with other users. Also you aren't promised speed parity with IPv4, because you aren't paying for a service. The tunnelbroker is to help those without native IPv6, get access. If only getting 40mbit/s from a free service is unacceptable, I think you would be better off spending your time yelling at your current ISP or finding another that can provide native IPv6 at the speeds you obviously find acceptable for paid products.

[roberttk]

I didn't say anything bad about tunnelbroker. I'm actually very glad to have a free IPv6 service.
I'm just trying to figure out WHY I don't get more speed. It is definitely possible, as other test results in this thread prove.
Also, I have connected a server to an IPv6 tunnel and I'm able to move data at 600 mbps (Amsterdam tunnel server).

I've run another test:
I'm using two entirely different IPv6 tunnels, one from HE and one from SixXS, on two virtual machines. Now when I start one download on each machine from a different test server, I'm still capped at just under 50 mbps in total:
<- taken from my router
This means it's not the connection to the tunnel.
I'm thinking that my provider throttles the tunnel connection somehow.
Any advice?

[broquea]

Same advice, talk with your provider. Find out if they are QoSing Protocol 41 below other traffic types. And find out when they'll provide native :)

[roberttk]

Officially they don't QoS anything.
They already provide IPv6 but only DS Lite and that's unacceptable for me.

There must be some way to prove they are limiting the tunnel's speed.

[simonszu]

Hi roberttk,

today i changed my contract with Unitymedia to the biggest one, 200MBit down, and 10MBit up. It seems that i have exactly the same issue, that traffic via the HEnet-tunnel is somehow capped to around 40-50mbit. Did you get a solution? Maybe another broker or something?

[tjeske]

If you're on DS-Lite, you should already have native IPv6-connectivity from your ISP. No need to use the tunnel at all.