Note:
Not all locations or servers support QUIC Obfuscation.
They did this with DAITA and it is annoying.
(DAITA is useless as a result because the few servers that support it are overloaded.)
I don’t understand why they are selectively enabling features on some servers.
Well, you can still get DAITA on any server - they just multihop. So, I don’t think its annoying.
No, that is unrelated, it still routes through a very small amount of servers, there are currently only 9 in USA for example.
How is it unrelated? You can literally get DAITA on any location/server, the app will just route you through it’s direct DAITA servers. What’s the problem?
I don’t understand your point.
Please read.
It is slow, because all users regardless of exit have to route their entry through a small amount of overloaded servers.
For stats: Mullvad has 209 servers in USA, only 9 of those support DAITA. That means anyone with DAITA enable and USA as their entrypoint (or exit if no multihop) is squeezed onto those 9 servers instead of spreadout across 209, causing extreme performance slowdowns.
Edit: global stats:
544 servers total currently online, only 48 currently support DAITA, so only 8.8%.
Assuming 10Gbps for all (some are more, and some few are 1Gbps): 5,440 Gbps total, only 480Gbps for DAITA, a small pipe is a slow pipe
That’s not what necessarily makes it slow. You please read.
DAITA, given the nature of the feature, will make the network speed slow anyway. Using multi-hop could make it a little more slower but on fast enough connections (which most internet connections these days are fast enough), it’s not noticeable for your everyday browsing needs.
Furthermore, I took issue to your characterization of DAITA being called
which is factually inaccurate because technically DAITA is available on any server (with oe without multi-hop but still available).
No, it has no inherent impact on peak TCP window size and resulting performance.
Typical performance with DAITA is single digit Megabits p/s, whereas without it is triple digits.
This is false.
I do not appreciate you spreading misinformation on this simple matter.
This is an extremely simple math problem: 209x10Gbps = ~2090Gbps vs 9x10Gbps = 90Gbps.
How do you think people can reasonably share 4.3% of their total USA server bandwidth and not experience slowdown?
If I can get DAITA on any server, with or without multi-hop, then how is what I am saying false. Please explain the logic with which you think I am wrong here?
Sorry, I am not seeing it your way.
This has NOTHING to do with compatibility. Please read what I’m writing.
It is about available bandwidth and slowdowns.
If it has nothing to do with compatibility, then why say I am wrong or spreading misinformation since it is available on any server?
You’re answering/addressing a different point than what I brought up and then saying I am wrong when you don’t even think it’s the issue that we are discussing.
I read. I understood. But you’re conflating two different things here then as you see it.
I never said there was a compatibility issue, just a speed issue.
It was implied (to me at least) when you originally said
–
That’s it. That’s all I was pointing out. I am not contesting your other claims but only that DAITA is indeed not useless because you can avail it anywhere.
Am I talking to a wall?
Do you not see the part where I said in my original post that it is overloaded.
Using multihop to get it anywhere does nothing about the fact that the pipe (available bandwidth) is very small due to limited actual server availability.
This overload directly makes it useless, because people will turn it off due to how slow it is, despite not being a technological limitation.
Yes, I read it. Overloaded. I get it.
But still not useless as you describe it. You can still use it. Also, all of this depends on your bandwidth and the location, server you connect to and whatnot. Many variables here for any one of us to have a singular claim about DAITA’s performance.
And there it is. This is the assessment with which I disagree. It’s not that slow (atleast in my experience).
We’re debating subjectivity/difference of experience here. Let’s stop.
I don’t think you can make this argument though since Mullvad doesn’t share their server loads. So it’s only speculation. In my experience DAITA servers have been just fine when browsing in Europe.
Mullvad doesn’t use 1 GBps servers for DAITA. All are either 10, 20 or 40 Gbps.
I can confirm the same for North America and the Gulf.
Yeah, no. When the only DAITA servers are halfway across the world from me, multi-hopping results in latency so bad I might as well just use TOR if I wanted to relive the “gool ol” dial-up days.
You’re free to take issue with however we judge Mullvad’s DAITA rollout, but at the end of the day, as customers, we also have an equal right to take issue with the dogwater performance.
Heck, I currently get better latency via NymVPN’s 5-hop mixnet compared to a Mullvad 2-hop DAITA connection.