Goodbye PowerDNS
blog.powerdns.com
blog.powerdns.com
<<DNS and the Cloud
Later, I saw that there was a push for “the cloud” to take over yet another part of our Internet. Encrypted DNS is great, we should all do far more of that. But I was (and am) tremendously unhappy that more and more of DNS is now set to move to (among others) Google and Cloudflare control – both of whom protest that they have nothing but the best intentions. But still I see yet more of the Internet getting centralised, and I worry where that will go.
I also worry that people somehow are not worrying about this – somehow we’ve made peace with the fact that companies far away get very detailed records on everything we do online, and that we just have to live with that.>>
Though, I should disclose, that this was in Germany, and for some reason I have found Germans to be particulary strong advocates of the decentralized web and general FOSS things.
With apologies to Mr. Godwin, I think it's simply that Germans have (in general) learned what is possible when data is centralized and administered in one place, easily accessible.
Additionally, I find an strange sense of pride (seeing as, again, I'm not German) that there are still some people who still live to fight all this data-collection and privacy invasion. I always look forward to, e.g. every year's C3 talks.
I frequently and publicly complain about crazy surveillance policy as much as the next guy. But I find it important to understand that the politicians in the interior and justice branches are not pushing mass surveillance because they're evil. It's because of perverse incentives: They know full well that, when the next terrorist attack happens, everyone who didn't push hard enough for law and order will be portrayed as an enabler. Or at least they think that this will happen. It can be argued if it really would, but them thinking that this is how the world is is my best explanation for their behavior. I don't think they want to build the next Stasi or Gestapo.
This, I'd argue, leads to it always being essentially career-wise risk-free to be more tough/stringent/et c. on <evil thing du jour>. And why not? There's hardly any downside. It's all unfortunately about optics.
EDIT: re-reading your comment, I realize I'm just stating in more words your salient point. It is, however, a particularly difficult hole to dig ourselves up from though.
https://www.theregister.com/2017/08/14/godwins_law_creator_r...
You might argue that the professors have no choice. But then there actually are non-CS professors from other universities that won't even put recorded videos of themselves online. They teach by uploading a few documents. Some stopped teaching altogether. None of that is particularly great. But it shows that professors have some power and would have even more of it as a collective. That in turn means that all those CS professors chose not to exercise it. From what I know it's the same at many other universities.
I seriously doubt anyone is taught any of the values you're talking about any more. The opposite is the case - if you held any of them before you'll have to be practical and let them go.
I've always thought a little history makes the lecture more interesting. Yet dry and boring lectures seem to be the norm.
Have you got specific examples? I can't recall this happening.
Some of it I agree with. Hate speech, disinformation, etc. but if google "owns" DNS than it's going to be the same story.
To take a specific service and say they haven’t is just naive.
Of course, this question would be silly under a dictatorship; the answer would just be "records have been suppressed." But we don't live in a dictatorship. Anyone in the US can use any DNS provider they like, even ones from outside the US. And anyone outside the US can use our DNS providers. So why would there be no evidence?
Mind you, I'm not necessarily arguing that Google wouldn't do this, if they were demanded to do so via a National Security Letter. I'm just arguing that you shouldn't conflate "no evidence that they're not willing to" with "already have in the past."
For IPv6 there is plenty of space and if you have a real use for more than a /48 then you can write up the paperwork for why you need so much space and get it approved, you don't need IANA, whether in Switzerland or anywhere else.
The only distribution is in the form of multiple levels of widespread caching. As long as that caching layer exists, is there really a difference between a large company hosting many names or multiple people hosting their own nameservers? Was the web really more resilient when everyone had their own servers?
I got to work a lot with PowerDNS during my internship, which was at a web hosting company that wanted to implement DNSSEC. What a fantastic piece of software it has become. Thanks for your involvement Bert, and best of luck at the TIB!
[1] https://berthub.eu/articles/posts/history-of-powerdns-1999-2...
I think it is great that he walks away when his hart is not in it anymore. I think a lot of big open source projects can become a big burden on its creators and some of them not daring to move on because of the feeling of responsibility. Same with the move of Redis’ founder it think these are great examples for the community.
We do not correlate or associate personal information in Google Public DNS logs with your information from use of any other Google service except for addressing security and abuse." https://developers.google.com/speed/public-dns/privacy
Do you have proof that the opposite is true?
Why would this be necessary with the big gaping hole in the assertion?
> except for addressing security and abuse.
I expect Google and I have very different conceptions of what qualifies as "abuse" and how to "address" it.
Seems like _not_ giving away data is the only way to be sure if that. Everything else is subject to change.
The "Google does not use any personal information collected through the Public DNS service to target ads" sentence is oddly specific as well. Why is it only mentioning "targeted ads" and not "for any sort of purposed"? Why is it collected if you're not using it? What does "abuse" really mean in this context, and what does "addressing" it entail? Only the big guys at Google knows.
Finally, the holy privacy policy of Google can change at any time. "We change this Privacy Policy from time to time. We will not reduce your rights under this Privacy Policy without your explicit consent." it says, which makes sense if you're a Google user. What about me, who don't have a Google account? How do I, tell this public DNS service that I don't agree to the new privacy policy? Right, I can't, you assume I'm fine with it.
I think we're long past the point of having to prove that various tech companies don't care about people's privacy. The burden of proof that you're actively taking steps to preserve it is on you, as you've already burned all the goodwill you had from the internet crowd.
"We take your privacy, very seriously"
Why are they collecting personal information in the first place, for a public DNS service that is supposedly built for "faster and safer DNS infrastructure"? Collecting no personal data would certainly be the fastest and safest for me as a user.
There is a huge distinction to be made in collecting vs. storing. The 8888 privacy policy also calls out that some personal information is stored, for how long, and for what purpose. But most people who claim that they're not collecting personal information are just not being accurate in their privacy statements.
Disclaimer: googler, opinion is my own and not my employer's.
What the hell are you talking about?! Of course you can, why wouldn't you be able to? You think the internet and the web started out with having everyone automatically collecting everyones personal information?
> Disclaimer: googler
Aha, yeaaaah, now it makes sense that you don't think it's possible to do anything else but to collect personal data. While the opinion is your own, it's pretty obvious what its been influenced by.
"You cannot run any internet service without collecting personal information" is not true, you can absolutely run services on the internet that collects and/or stores 0 personal data. That they don't know how to, is besides the point, but that doesn't make it impossible, it just means they don't know how to.
Now you can probably keep this for typical security purposes for a week or so, and even aggregated page hits, without permission so you're not entirely wrong
Running netflow probably counts as collecting PII, but the main idea is
1) Why are you collecting it. Clearly storing connection details in a firewall is acceptable.
2) Why are you keeping it. Storing IP logs for security purposes is fine for example.
3) How are you keeping it secure and ensuring you only keep it for the shortest amount of time you need.
Yeah, and the defaults for mongodb is to let anyone edit the database, at least it used to be so. Doesn't mean you should run it like that.
You can, as soon as you receive a request, hash the IP address and store the hash instead of storing the cleartext IP. If there is no way of reversing the hash (which there shouldn't, wouldn't be much of a hash then), you're not storing any IP anymore, and it's not personal data.
Strikes me as very weird someone who supposed to be high-level developer or employee at one of the biggest companies in the world don't know how not build applications that stores personal information. But since that company is Google, it hardly surprises anyone.
There's nothing wrong with storing individual IPs temporarily for security and functionality purposes, but you're still collecting them - if nothing else they'll show up in your netflow sessions and your netstat output
I didn't know! Here is my IP as a sha256, please reverse it and show me you actually know what you're talking about: 20adfe5288d85b6279dc65eff126828c0b20b896cb853caef9de0d6251e46169
This quote is from 2012: "I do 622 million SHA-256's per sec on a Radeon HD5830."
You might have salted it or encoded in a 'clever' way like hex or words or whatever, but as you know the method you can do it.
What prevents you, in future when policy changes, to start using such data?
Even in cases where using the data would directly violate a contract, bankruptcy courts / acquisition agreements routinely sell the data off as an “asset” and nullify any restrictions from existing agreements.
A change in the law that kept the copyright of personal data with the individual the data was about would improve the situation somewhat, but it’s probably safer to just ban the collection of this sort of data set (unless there’s explicit opt-in, and people derive no benefit from opting in).
Google has simply shut down too many information channels (MMS, RSS, Usenet, forcing https, to name a few) in the name of "security", when they are the ones to gain from this deprivation and monopolization into the "open web".
Nobody can convince me Google has oppressed political minorities in mind with DNS-over-HTTP.
All those products that got killed mostly started that way, employee's wanted them and google had the money to make them happen. Eventually though if they're not growing, then who in google wants to work on them?
Did google also see it as a strategic way to make sure dns providers couldn't force search results elsewhere? Probably, but then as users WE certainly don't want dns providers fucking with dns in that manner either.
Google wins when the Internet is fast. Google's goals here are that you can use Websearch quickly and without hotspots etc meddling with traffic. The story is simple: If companies meddle with DNS results, or provide poor performance, then users don't use the Internet as much, and Google doesn't get to sell ads. Google wants you to have the fastest, most reliable Internet it can, because that means you don't get frustrated and go do something else. Google doesn't need to scrape the logs for whatever people think google might be scraping the logs for.
Google _does_ log some data for dealing with abuse (as you can probably imagine 8.8.8.8 gets a _lot_ of intentional and unintentional abuse), but tries to be as clear as it can be about what gets logged, what it's used for, and how long it's kept. Google treats these logs very very carefully, carefully limiting who has access, and when you need to use the logs to debug something, it's very carefully audited.
Disclaimer: I used to be one of the SREs oncall for Google Public DNS (but not any more – I now work for a different SRE team at Google).
IMO, it is reasonable to assume that an advertising company would use data (that consumers give them) to further enhance and target ads to those consumers.
yet.
And Goolge is way past where I assume they have good intentions. Do you have any proof Google isn't doing something creepy with that data it collects from DNS logs ?
For each 12 TB they purchase for logging DNS queries per day, they could store a bit over 2KB per user. Assuming reasonable compression, they shouldn’t be spending more than 2-4B to log each request, so they can store 500-1000 queries per user.
My guess is they could store everything, but choose to apply some aggregation.
“We collect personal information through the public DNS service, but do not use it to target ads.”
Sentence 2: We log dns queries and correlate and associate personal information in these logs with other google and third party (and presumably Alphabet, YouTube, etc) services.
Datasets that were correlated from Google services is only used to address security and abuse.
No, you changed the meaning when editing. The original sentence does not confirm the personal data is collected, but if it is it's not used to target ads. This is pedantic, but if you're trying to explain someone's written policy it's important not to put words in their mouth.
Would you sign a lease that said the landlord wouldn’t use any video surveillance from hidden cameras inside the home to bring legal charges against the tenants?
You can be sure every word of this part of the privacy policy was crafted to sound as reasonable as possible without limiting any of their existing behavior.
I have a vanity domain which I use to exercise my internet independence :) I've been using PowerDNS with a hidden master, public secondaries, dns updates with lua authorization for dynamic ips and let's encrypt for years. Also the recursor for local systems.
I received great help and had some fun chats on IRC in the beginning. PowerDNS has been a great part of my "infrastructure". My only complaint is that packagers and other naming can't choose between 'powerdns' and 'pdns' :)
What a ton of people don’t understand is that you don’t need a DNS forwarder (google/cloud flare/ISP).
If you setup your DNS right it just goes to the ROOT servers and you cut out those parties spying on DNS traffic.
Initial requests are a bit slower because you don’t benefit from the forwarder’s cache, but after a while, most stuff is coached and it’s not a big deal.
Maybe I’m totally wrong here, but this is how I understand it.
Honestly interested here.
https://www.internic.net/domain/named.root
This Ansible role takes care of configuring it all (CentOS):
https://github.com/liv-io/ansible-roles-centos/tree/master/u...
I have it for FreeBSD and OpenBSD too:
https://github.com/liv-io/ansible-roles-bsd/tree/master/unbo...
I have been using this program on all my computers, during the last almost twenty years, without problems.
DNS is very easy to run yourself. I use, and highly recommend, Unbound[1] as a caching resolver, and NSD[2] as an authoritative nameserver.
A caching resolver is used for ordinary lookups. If the address for "ycombinator.com" isn't in the cache, the resolver starts at the root and recursively finds the authority for the domain, gets the answer, caches it, and returns it to you. You run this for general usage, and you point your desktop/laptop computers to this address. Google runs their public resolver at "8.8.8.8", etc.
An authoritative nameserver is the ultimate authority on the lookup results for a specific domain. All lookups by other people eventually reach the authoritative nameserver for an answer. You run this for domains that you own, and configure it in your domain record.
(Both can be a bit more complicated than that, but they don't need to be)
These two services need almost zero care and feeding. Set them up once, and you can basically forget about them for years. This article only discusses resolvers, which are literally set-and-forget. Authoritative nameservers only need maintenance if you make changes to your domain. (And if you don't have your own domains, you don't need an authoritative nameserver at all).
[1] Unbound: https://www.nlnetlabs.nl/projects/unbound/about/
Totally agree there.
Root DNS servers would not answer your computer's (stub resolver's) _recursive_ query. You do need a recursive resolver to do that - whether ran by an ISP or by yourself.
No matter what you do locally (unless you locally serve the do-not-auto-doh flag), you'll have to touch each of your browsers on each system to not have this behavior.
While not exciting for you anymore, I am happy that it continues to run without much excitement.
Thanks also for making sure it is in good hands after you step down, much appreciated.
Wishing you the best on your new endeavors. Keep life exciting and live it to the max :)
I would love to see it updated to us em instead. Since I actually have tweaked my browser settings to by default use a larger font that statically defined pixels don't listen to.
Once I zoomed the blog to 125% I was able to read it just fine but read it at 133% since it was more comfortable for me.
* https://twitter.com/PowerDNS_Bert
He also has:
That became unworkable about three and a half decades or so ago, however, and the DNS was created to solve that problem.
Assume each zonefile is 1Kb that could be 350GB of data, right?
https://blog.powerdns.com/2019/09/25/centralised-doh-is-bad-...
https://berthub.eu/articles/posts/history-of-powerdns-1999-2...