Chromium cleans up its act and daily DNS root server queries drop by 50 billion
- Reference: 1612425664
- News link: https://www.theregister.co.uk/2021/02/04/chromium_cleanup_drops_dns_traffic/
- Source link:
As The Register reported in [1]August 2020 , Chromium-based browsers make a lot of DNS traffic as they try to determine if input into their Omnibox is a domain name or a search query.
[2]
Verisign engineers Matthew Thomas and Duane Wessels examined the resulting traffic and reached the conclusion that it accounted for up to 60 billion DNS queries every day.
Wessels has since penned a new post that went unreported when it appeared on January 7th – the day after the US Capitol riot – but was today resurfaced by APNIC, the Regional Internet Registry for the Asia-Pacific region.
[3]
All too often, technologists solve problems by introducing additional layers and disregarding simpler solutions
In the [4]post he says the Chromium team redesigned its code to stop junk DNS requests.
The result? “Before the software release, the root server system saw peaks of ~143 billion queries per day,” he wrote. “Traffic volumes have since decreased to ~84 billion queries a day.”
“This represents more than a 41% reduction of total query volume.”
Wessels post praises that outcome, but also offers some lessons for the future.
Ever wanted to own a piece of the internet? Now you can: $1 for a whole gTLD... or $2.8m if you want a decent one [5]READ MORE
“All too often, technologists solve problems by introducing additional layers of technology abstractions and disregarding simpler solutions, such as outreach and engagement,” he wrote.
[6]
“Chromium’s efforts show how such outreach and community engagement can have significant impact both to the parties directly involved, and to the broader community. Chromium’s actions will directly aide and ease the operational costs to mitigate attacks at the root. Reducing the root server system load by 41%, with potential further reduction depending on future Chromium deployment decisions, will lighten operational costs incurred to mitigate attacks by relinquishing their computing and network resources.” ®
Get our [7]Tech Resources
[1] https://www.theregister.com/2020/08/21/chromiums_dns_network/
[2] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_datacentre/networks&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2YBvTz-eUsa@qSKwqKoNxngAAANM&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_datacentre/networks&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33YBvTz-eUsa@qSKwqKoNxngAAANM&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[4] https://blog.apnic.net/2021/02/04/how-chromium-reduces-root-dns-traffic/
[5] https://www.theregister.com/2021/02/04/uniregistry_domain_auction/
[6] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_datacentre/networks&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44YBvTz-eUsa@qSKwqKoNxngAAANM&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[7] https://whitepapers.theregister.com/
Let's just use lots of public resources
to make shiny. The commons belong to us!
hang on
So google merge their search and address fields into an "omnibox" and it makes billions of unnecessary DNS requests. Now they've calmed it down a bit, that's an example of how great they are at engaging with "the community"?
Shouldn't they have done the community engagement thing before merging the fields?
Re: hang on
I like the merged box. If nothing else it's simpler to explain to the mundanes. The fact that their initial implementation sucked is another matter.
The fact that is sucked for somebody other than themselves is of course why it stayed too long.
Re: hang on
WHAT? You mean design the thing and look for possible inefficiencies and cockups before building it? It'll never catch on.
Simpler to just chuck together a prototype, lob that into beta, fix the serious user gripes and chuck hardware at any performance issues.
Costs so much less and makes the development process more......erm......agile. Yes. That's a good word, agile. Makes it sound like a method rather than just a massive fuckup looking for a place to happen.
Re: hang on
It's one of those irregular verbs.
I'm agile. You parrot buzzwords. He does a random walk.
Re: hang on
Yes, thank you, Bernard.
Re: hang on
Also, why were they hitting the root servers? Shouldn't it just be using whichever DNS the machine it is on is configured to use?
Disclaimer: I only have a basic understanding of the DNS infrastructure so there may well be a good reason for this!
Re: hang on
If I'm searching for a term, say "widget", it's probably trying to find widget.com, widget.org, widget.co.uk, widget.net and a whole bunch of other domains. Yes, those queries will go to my "local" DNS server (which could be a corporate one or my ISP). Unless they've cached that information already, they need to look to the root servers to find the answer, hence it hits the root servers eventually. They should cache that for future requests by the next person searching for those domains reducing future traffic, though.
Re: hang on
And also search for the TLD w, then wi, then wid, then widg, then widge, then widget
Re: hang on
Any query that can't be resolved locally will hit the root servers for a definitive "go here and ask them". Obviously "corking cakes" never resolves and it probably makes a request for each letter.
Re: hang on
Bingo! Is the send every fucking letter to google one at a time in packets of 1k...its the ultimate amplification attack, so they can send you back matches for each keystroke, lookup up the resultant cacaugphony (sp?) every time. This plus their 'generate random string and query it to determine if dns is being hijacked'
The only reason they fixed it is because, today, even Google employees are working from home and no longer have 1G-100G sub-millisecond connections to their servers. I would love to have been able to listen in on the bitch sessions for every time someone said "but it worked just fine on friday in the office, not its not responding at all"... serves you right, bastards.
Re: hang on
As far as I understand, it will speak to the local DNS, but if it doesn't recognise the domain in its cache it escalates it 'to the manager' who knows everything.
Re: hang on
It's not that simple. If my local DNS doesn't know the answer, it doesn't just go straight to root. It goes to another level (often the ISP DNS). If that doesn't know the answer, it will look elsewhere, eventually getting to root. Any organisation that goes straight to root is being very rude.
well how about...
If the user starts an entry with http(|s):// or even www. or ends an entry with .blah, it gets treated as a URL and triggers a DNS search.
If not, the query goes to the search engine of choice.
One might reasonably expect that Google actually have a list of URLs that could be tested for before troubling the DNS servers, no? But that would be using Google's hardware rather than the commons, and perhaps they don't like that.
Re: well how about...
No. I switch off every option to have my URL/search terms/whatever sent to be indexed by the surveillance machine. (At least, every one I can find.)
Re: well how about...
I think you'll find Google's public DNS is at 8.8.8.8 and knows pretty much everything. I'm surprised it ever has to go to root.
There was never any need for the excess root server DNS queries. It was just lazy programming. For Google or the Chromium project to claim otherwise is disingenuous to say the least.
Truer words...
“All too often, technologists solve problems by introducing additional layers of technology abstractions and disregarding simpler solutions, such as outreach and engagement,” he wrote.