Live data from Hacker News

VPNs are not the solution to a policy problem

asininetech.com

31–40 of 228 posts

Re: VPNs are not the solution to a policy problem

#31
Technology used to trump policy, in an unstable but stubborn way. Napsters and piratebays die, but file sharing lives. It's less intense now nit because of policies, but because legal ways to buy most music and videos became reasonably convenient for the mass user.

How well might connectivity limitation work? It took China immense centralization and a lot of technical effort to build the great firewall, which is not exactly impenetrable, though.

Re: VPNs are not the solution to a policy problem

#32

A heads up: theres a really nice project called Streisand[1] which provides a multi-protocol VPN with very little effort. You can launch one on a cheap cloud provider (like DO, if their policy allows). [1] https://github.com/jlund/streisand

Any estimate on EC2 costs using this moderately?

Re: VPNs are not the solution to a policy problem

#33

A heads up: theres a really nice project called Streisand[1] which provides a multi-protocol VPN with very little effort. You can launch one on a cheap cloud provider (like DO, if their policy allows). [1] https://github.com/jlund/streisand

I've used streisand on DO (while traveling in China) and it worked well. There's also a similar project called algo[1] which provides a single protocol with maximum security, in contrast to streisand's multi-protocol flexibility (and increased surface area).

https://github.com/trailofbits/algo

Re: VPNs are not the solution to a policy problem

#34

I think the bigger hole is DNS. Full-tunnel VPNs to primarily TLS-encrypted sites seems like overkill. Encrypted DNS plus an "HTTPS Everywhere" plugin should obfuscate enough info for most people without significantly affecting latency.

DNSCrypt + HTTPS everywhere solves the latency issue but it doesn't solve some of the other issues.

You still need the technical know-how to set up a DNSCrypt recursive resolver. The resolver then talks to the respective recursive chain in plain text as DNSCrypt is not something that is widely adopted.

Re: VPNs are not the solution to a policy problem

#35
post #19

I think the bigger hole is DNS. Full-tunnel VPNs to primarily TLS-encrypted sites seems like overkill. Encrypted DNS plus an "HTTPS Everywhere" plugin should obfuscate enough info for most people without significantly affecting latency.

Wouldn't it be fairly trivial to guess most of the domains you're visiting by looking at what IP addresses you connect to?

Yes. To be fair though, many sites are on shared hosts, and lots of traffic goes through a handful of CDN networks.

I think that the SNI note below is probably the bigger hole.

Re: VPNs are not the solution to a policy problem

#36
Another thing often overlooked with VPNs is that they're just not that fast. I have a 600/40 connection, and I've tried at least six for-pay VPN providers. The fastest one I found (won't mention as my goal isn't to advertise for them) hits, at best, 100/30. And even then, only over L2TP. For whatever reason, OpenVPN is always slower on every PC I've tried this with.

And obviously, you gain a good deal of latency, especially if you use an overseas exit point.

And now we get to deal with shitty services like Netflix punishing privacy-conscious users and blocking access to paid accounts while your VPN is up.

Re: VPNs are not the solution to a policy problem

#37

A heads up: theres a really nice project called Streisand[1] which provides a multi-protocol VPN with very little effort. You can launch one on a cheap cloud provider (like DO, if their policy allows). [1] https://github.com/jlund/streisand

I've used streisand on DO (while traveling in China) and it worked well. There's also a similar project called algo[1] which provides a single protocol with maximum security, in contrast to streisand's multi-protocol flexibility (and increased surface area). https://github.com/trailofbits/algo

Why does he refer to OpenVPN as a "risky server"? Does it have a history of embarrassing security vulns?

Re: VPNs are not the solution to a policy problem

#38
I had all sorts of VPN problems over the years with various Linux desktops OS. What I do instead is that I have a proxy server with just an OpenSSH daemon on port 443 -- if there's web traffic, add sslh to taste -- and then use the SOCKS v5 proxy built into OpenSSH client and then http://darkk.net.ru/redsocks/ I might be the weird case here but I found this infinitely easier to set up than any VPN.

Re: VPNs are not the solution to a policy problem

#39

Until a better solution is found, I think the way the recent IOT botnet stuff + this ISP privacy deregulation is portrayed in the media opens the opportunity for a startup that sells a secure, smart home router + VPN subscription plan.

And you would trust the startup with your traffic because...?

End to end encryption, keys never leave your premises, routing is randomized? Looks a bit like TOR.

Re: VPNs are not the solution to a policy problem

#40

I think the bigger hole is DNS. Full-tunnel VPNs to primarily TLS-encrypted sites seems like overkill. Encrypted DNS plus an "HTTPS Everywhere" plugin should obfuscate enough info for most people without significantly affecting latency.

DNSCrypt + HTTPS everywhere solves the latency issue but it doesn't solve some of the other issues. You still need the technical know-how to set up a DNSCrypt recursive resolver. The resolver then talks to the respective recursive chain in plain text as DNSCrypt is not something that is widely adopted.

Hosting a private DNS server has its own issues. Many CDNs rely on DNS server to determine which POP to route you to. Pretty common for Australian internet users who switch their DNS to have videos streamed from Southeast Asia rather than Australia. That would cause huge perceived latency issues. Third-party DNS providers solve this with private agreements [1].

[1] https://community.akamai.com/docs/DOC-4219

Post reply on HN