I also like that you don't have to give them any private info at all to make an account. You can just send crypto and they'll give you an account code and that's it, you don't even need an email address.
I haven't tried it but apparently you can even mail them cash. You get a payment token and just send cash in an envelope and they'll activate it whenever the money shows up!
Be aware that Mullvad recently removed support for port forwarding if that matters to you. They're no longer a preferred option for torrents for that reason. Other than that I enjoy using their service.
Longtime Mullvad user, always been happy. But when Mullvad was still a small service it was unusual to have any problems when browsing the web with their IPs.
Recently, many services can detect you're on a VPN when using Mullvad and block or ban you, which means they've become successful enough that there are countrer-VPN databases including all of their IPs.
Pretty sure fextra just rips all their content from other wikis anyway, at least this was definitely my experience in the past. Just try scrolling past the first link in your search engine.
I've just come to accept that constant captchas are a fact of life for browsing on a VPN. Cost of doing business. Worth it for the privacy though imo (VPNs in general, I haven't used Mullvad).
Some are definitely better than others. I've used new VPN services that get you through every checkpoint just like a home IP address. And some that, as you mention, throw up every captcha known to man.
The result is that the operating system that we boot, prior to being deployed weighs in at just over 200MB. When servers are rebooted or provisioned for the first time, we can be safe in the knowledge that we get a freshly built kernel, no traces of any log files, and a fully patched OS.
I assume they mean there are no account credentials. When you "create" an account on their website, you'll be given a random account number, and no password.
As they outline here, there are ~9 quadrillion possible keys, needing around 5.5 million guesses to find an account. I think they hit a nice middleground between decent entropy and still having a number you can memorize (like a credit card).
To be fair, would it matter if someone got access to your account key? There isn't really any data on your account is there (isn't that the point)? It'd just let you connect to the VPN
They can use your secondary connection for free. It depends if that bothers you or not. If you're already using both it could lead to disruption on your part I guess? Not 100% on that though
If the computer is powered off, moved or confiscated, there is no data to retrieve.
We get the operational benefits of having fewer breakable parts. Disks are among the components that break often. Therefore, switching away from them makes our infrastructure more reliable.
The operational tasks of setting up and upgrading package versions on servers become faster and easier.
While mostly true, there are ways to preserve ram if the device is confiscated.
Your local PD likely couldn’t pull it off, but if one of the larger abbreviation agencies were to get involved, data on RAM isn’t a huge hurdle. Assuming no one flips the power switch, at least.
I'm not an expert but I think : The site you visit only sees the VPNs info. Which is how you maintain some anonymity while browsing. However, if your VPN keeps logs, then you can still be tracked, just at a different place. Some say they don't keep logs, and you'd have to trust that.
RAM is considered volatile memory, so each time the server turns off, it loses all data. This is compared to disk (hard drives of whatever type) which retain memory even if the server turns off.
In theory, this ram only server prevents them from keeping logs (like which user went where) since the server wouldn't even have a place to store it.
Edit: lustrums post is more accurate and has info that this doesn't prevent logging per se, but could prevent accidental logging. I.e. they can't hire a forensic computer specialist to parse through operating system logs to try to find info they didn't otherwise log elsewhere.
Some say they don’t keep logs, and you’d have to trust that.
Note that this same caveat applies for a VPN provider promising they are running diskless endpoints. Or that they don't have some third party monitoring their stuff even if diskless. Or that a law enforcement agency can't come along with a warrant to require them to monitor an account's activity moving forward, even if logs are not possible.
If your online activity justifies this level of paranoia, there's probably no meaningful protection available for your wants in practice. If your provider is operating in a jurisdiction that is problematic for your online activity, they can probably ultimately be compromised. If you are just using it to access a different country's streaming library, you probably don't need to be that paranoid. If you are trying to disguise illegal activity that is illegal in the jurisdiction of the VPN endpoint, well you are likely boned with logging or not.
A normal computer is usually constantly writing little bits and pieces of data to disk. But data on the disk might accidentally remain on the disk even if it's not intended. Then that data could be read later by someone else who is spying on VPN users .
There's also a common assumption that data on disk storage may leave behind remnants even after it's been overwritten. (Magnetic disks may leave behind some magnetic signatures. Flash drives will stop using sectors that are worn out, potentially leaving data there.) And state actors like NSA might have some capability to recover this ghost data if they get a hold of the actual drives.
There's a general understanding that data on RAM is irrevocably destroyed within a short time after the device loses power. So attacks on RAM data have to occur in real time while the data is in use. (There may be some attacks that preserve RAM after power down using low temperatures and liquid nitrogen).
It means that even though Mullvad already doesn’t log anything about their users activities, there is no persistent storage on the servers, so as soon as it is powered off or raided by The Agencies, there is absolutely nothing to retrieve from it.
If the computer is unplugged, there's nothing left on a hard drive to show what state it was in. This means nobody malicious can physically remove their servers and gain information about customers.
Just for my understanding when they boot such a server, where does it get it's operating system from? Over the network from a different computer which has a hard drive or some read only ROM on the server or what?
You can boot from a HDD and then just not ever write data back to it. This would be the most trivial solution, and it's something people do with their Pi's a lot to avoid SD card failure.
You could network boot, pull the OS from the network at startup. Fun fact, this is how some rockets fly! No onboard persistent storage needed. Everything boots into and runs from ram the whole 10 ish minutes of operation.
You COULD do a ROM as you suggested, but that's a LOT of ROM. Seems odd to do imho.
I remember that there was a ROM in the Amiga 500 which had the kickstart software on it which you'd load from a diskette on the predecessor the Amiga 1000. This made it much faster to boot because you would not need to switch diskettes in the middle of the boot.
Click the first link in the article, in the older post they talk about their stboot bootloader. It does what you suspect, loads the OS image from a different computer which has signed base images.
Anyone pro-Mullvad that can explain to me how it's better than PIA?
To my knowledge, which may be wrong, PIA has faster speeds and is also entirely RAM-based.
That said...I'd gladly switch if that's untrue and Mullvad is better. On the outset, it sounds like Mullvad triggers search engine captchas less, which would be a nice win.
edit: Well, you all convinced me. Made the switch.
PIA and Mullvad should have equal speeds because they both have 10gbps servers and wireguard. Both PIA and Mullvad use ram-only servers exclusively. As for search engine captchas, I never get them with Mullvad. The main issue with PIA is that they were bought by a questionable company that previously developed adware. You can read about that here. Personally, I would never use a privacy tool that is owned by an ad company, even if they claim to have changed. I used them up until the acquisition, then switched and have been extremely happy with Mullvad.
As for search engine captchas, I never get them with Mullvad.
That has nothing to do with VPNs, and everything to do with how your browser “leaks” your user behaviour history.
Captchas go through your browser behaviour history and examine the clicks and pages you have gone through, how long you were on each one and how you scrolled through each page. Stuff like that. If that browser behaviour history reaches a minimum threshold of “human-like behaviour”, there is no test to pass. If it doesn’t, or there is no history to go after, you get a test.
As someone who works in enterprise ISP tech space I always keep the bigger picture in mind, especially with the latest "tech Fads", VPNs are really easy to sell, especially when you already have other companies and even bigger shell companies.
Take the following scenario (it might be true it might also be conjecture):
person1 owns 2 shell companies that are big names in tech.
shell 1 starts out as a an ISP and soon grows to be a network transit provider.*
shell 2 starts out as a cyber sec company.
shell 1 get's really big and becomes a tier 1 provider that sells transit to BBC and is now peering with the likes of Cogent, Lumen/CenturyLink and others.
shell 2 get so big it branches out into VPN carrier tech and purchases a well used VPN company that also stands out as having a no logging policy.*
shell 1 starts providing seriously detailed analytics to it peers on a subscription basis with discounts to peers that repeatedly hit the 95th percentile on billing cycles, all the peers love being able to see detailed info of the traffic flowing over their transit relationships.*
Shell 2 also purchases another company that deals with adware and advert injection tech.
later shell 2 becomes so financially liquid it is now breaking out in to gambling and lucrative AIM ventures.
In the scenario above I've marked points with a * that should be red flags to VPN users BUT they have something obvious when laid out in this manner that a user of a VPN would not know. That is that even though the VPN is sold as no-logging the wider company still gets your data as all the traffic is flowing over the wider network owned by shell 1 that you have no idea of the relationship between them.
All traffic/data can be monetised and ultimately with decent visibility of all comprising parts tied back to you or your account, VPNs are good but just be aware of forced perspective, look beyond T&C's, look at the company and who owns it and what else they own.
You all got a hint at this with pirate bay, the feds couldn't take 'em down so the went to the DC provider and the network transit providers, you should do the same if you value your trust and data so much that you need a VPN for every connection.
Finally, with or without a VPN, Your IP is only used for 20% of the connection(10% at the start and 10% to the final endpoint), when your data/traffic flows over provider networks it becomes an AS number, a layer tag and even a colour, all of these interchange until it becomes an IP again, hits a website and for the most part all of that is accounted for and can be connected to you.
Self hosting would essentially just be using a ramdisk. If you want to be crazy about it, you could even run a VM with its storage entirely within a ramdisk.
Note that the lack of logging probably doesn't matter when your self-hosting, since it's all for you.
Concept of RAM only Linux images with validation and signing is something seen in some datacenters. For example, Lenovo has this in their confluent cluster management (https://hpc.lenovo.com/). A node can network boot or boot from usb (read-only) and all writes go to RAM.
Alternatively, booting a LiveCD amounts to the same thing without requiring a boot server, you have a local 'disk' but nothing writes to it. If extra paranoid you could actually boot it from a burned DVD, but in practice even when booting from USB most 'live' images only write filesystem to RAM.