Hacker News Reader

Comment by 0daystock | Hacker News Reader

> Not only is it our policy that we would not accept logging, but we have also specifically designed our VPN servers to not be able to log, including by running in RAM.

Do people really believe this bullshit? Empty claims of servers running "in memory" as a meaningful defense against surveillance?

View on HN

Going diskless is not a complete defense (nothing is), but it still helps against certain attack vectors. Borrowing from Mullvad's blog post on the topic[1]:

- If the computer is powered off, moved or confiscated, there is no data to retrieve.

- Running the system in RAM does not prevent the possibility of logging. It does however minimise the risk of accidentally storing something that can later be retrieved.

https://mullvad.net/en/blog/2022/1/12/diskless-infrastructur...

But VPNs already promise not to log anything, so running in RAM makes about as much sense as "double encryption" (actually I think some VPNs do offer such a thing, and people are gullible enough to accept it).

Non-persistence of any data is a positive in terms of data at rest, how is that not a defense against surveillance? Regardless of the fact that it's not verifiable, assuming it were true, would it not be a good thing?

How is an anonymous, non-attributable, non-verifiable statement, from a company trying to sell you a product, worth anything? Why assume it's true when it is so contrary to even basic common sense, for anyone who has ever stood up a LAMP stack?

I don't mean to assume it's true to validate their advertising. For the sake of argument, if a server uses only a RAM disk, is that an improvement over having disks? Of course a network can still have sneaky equipment in between but that is still possible without RAM disks, so is it not beneficial to have a RAM disk?

Please note that I am not arguing in favor of their advertising or to say that it is successfully avoiding surveillance. But, do you believe that no-disk boot, assuming it actually takes place, is a positive thing, or not?

I admit I misconstrued your original comment to be a criticism of the technology rather than the fact that this VPN company advertises as such. Nonetheless I do think no-disk-boot is not useless as a technology and if you have any disagreement I would love to hear it, as someone who uses a VPN (not expressVPN) that says they use the same setup.

It depends entirely on the threat model. If three-letter agencies are the adversary, moving logging to RAM is unlikely to be a meaningful deterrent - they probably already have a root shell or direct access to the VM hypervisor. So if it doesn't protect against nation states, whom are we defending against? Another law enforcement agency raiding the server room and taking hard drives? But I thought ExpressVPN doesn't log anything, so why would that matter? Let's just be real and practical about what problem this is actually purported to solve, else we should call it theater because it's what it is.

the question was, is it better (than running off-disk), rather than "is it good enough for X"

the answer is, yes. even an infintesimally smaller attack surface is better than an infinitesimally larger one, all other things being equal

It's not at all better - it's worse, exactly because it gives a false sense of security.

you claim to know it gives you a false sense of security -- this knowledge itself of the possibility allays the concern

Because it's pretty easy to set up a ready only server that runs off read only storage and that doesn't include any writable storage. The fact that it could be done doesn't really require extreme proof. They could still be lying but it's not a hard or unusual thing to do.

these claims aren't necessarily empty, but pointless because ISPs still log everything, as they're required by law pretty much everywhere. it would require a bit more digging to through, but the data is still there - even with some clever routing with on-premise equipment, there's still more than enough data to deduce which inbound connection corresponds to which outbound connection

if your threat model is three letters agencies, vpns and tor are a fig leaf

I agree a VPN won't help against a three letter agency. But it will help against an ISP, who has a legal right to sell your browsing data in the US.

This is one of the use cases for why you might want a VPN, if you trust a VPN company more than your ISP.

A VPN is just paying for putting your trust in a VPN brand rather than an ISP brand. I don't see why that's such an offensive business to so many HN users.

> I don't see why that's such an offensive business to so many HN users.

Because the assertion VPNs - apparently unlike every other ISP - do not log or monetize your data is simply laughable, especially as so many are based in third-world countries, set up by shell entities and have almost no accountability for any of their claims.

Their entire business model is premised on the fact that they don't. If they ever were found to be, their hundred million dollar businesses (expressvpn) would vanish.

When express has their servers seized in turkey, there was no usable data on them.

I know you are super paranoid, but that still doesn't make my point wrong, or using a VPN wrong. Again, if you trust a vpn more than your ISP, that's pretty legitimate in many countries.

Your trust is based on baseless promises and one isolated, possibly even manufactured, incident. I find this to be super naive.

You might as well just treat everything as manufactured then. Even this thread must just be manufactured by Big VPN.

Not an unreasonable thesis in the age of misinformation.

You clearly have 0 trust in your life. Might want to work on that.

Mostly only for untrustworthy VPN services that try to make a buck on technologically illiterate audiences.

Well then maybe look into the practices of your ISP as well. I don’t see why you trust them more.

I'm as generally skeptical as anyone, but I think you've seen that it's essentially impossible to rationally debate with someone who believes in a conspiracy to the point that evidence against it can just be dismissed as part of the conspiracy itself. It's frustrating.

I certainly wouldn't trust my life to an unaudited VPN, but I think your two main points are pretty compelling -- 1) the business model is of large VPN companies is based on trust. They have very explicit, business interest in not violating that trust. 2) in one case we know of, when seized, the servers didn't have actionable information on them.

Does that mean every VPN company is trustworthy? Of course not. Does it mean that things could change at any time? Of course.

Microsoft used to crow about this stuff a lot of with respect to O365. I remember getting a dirty look when I laughed at the rep.

Their services terminate TLS locally for most tiers of service (Even with the “Government Community Cloud”), so you need to be careful and use VPNs in any scenario where a foreign interest may be interested in what your employees are up to.

Reply on news.ycombinator.com