3 ms·

Because you made a zero effort joke that has been worn to the absolute bone by people repeating it thousands of times per year in HN comments, rather than at le

by jsnell 6y ago

Because you made a zero effort joke that has been worn to the absolute bone by people repeating it thousands of times per year in HN comments, rather than at least putting in a bit of thought as to what the risk is for this service. Here's a quick attempt: Track record: 8888 has been running for more than ten years. Popularity: It appears to be incredibly widely used; e.g. a research paper from 2013 claimed it was serving 7% of all end user DNS queries, and Wikipedia claims that in 2018 it answered a trillion DNS requests per day (i.e. 10M qps). Business value: It is true that there is no direct revenue here. But I'm pretty sure that the original reason for launching it was defensive. A lot of Google's networking projects have clearly been driven by trying to ensure that users can connect straight and reliable to Google's services with no interference from middleboxes, since every request lost to interference is also lost ad revenue. Crappy ISP DNS servers that serve spam pages instead of NXDOMAIN are a pretty big vector here.

willis936 6y ago
It’s not a joke. It’s a question of trust and authority. “Google DNS has been running for 10 years” does not negate Google’s management behavior or change the inherent risk of a free running for-profit entity being considered a DNS authority. You calling it a tired joke doesn’t change these things.
acdha 6y ago
On the business value front, I’d also bet that they see value from the data visibility front since that gives them a chance to identify malware and phishing domains early and the terms of service appear to allow aggregate activity data mining. It’s not as precise as Google Analytics, of course, but it works on many things which don’t have JS beacons and I’d be surprised if they couldn’t find a good deal of useful information from that stream.
drewbug01 6y ago
> Track record: 8888 has been running for more than ten years. I understand the point you're trying to make here - but long-lived products are still shut down by Google, fairly often. Here's a list of some Google products that have been / were around for 10 years or more before being killed (or are slated to be killed soon): - Chrome Apps - Cloud Print - Fusion Tables - Youtube Video Annotations - Google Search Appliance (17 years old!!) - Google Showtimes - Google Code - Picasa - Orkut - Postini That's not even close to the full list of things that were more than 10 years old when Google shut them down. I left a bunch off for brevity's sake - but browsing https://killedbygoogle.com https://killedbygoogle.com is a really eye-opening experience. They really have shut down a lot of stuff - and if you loosen the requirement to ">= 7 years" the list gets very, very long. Again - I understand what you're trying to say. But it's just not simply a "zero effort joke". Google has killed a lot of things, and a lot of the things they killed were well used, long lived, and popular. They'll keep Google DNS running as long as it provides business value to them, and that's it. (Edit: I'm bad at list formatting, sorry)
- colinmhayes 6y ago
  Use your brain. Google kills product because no one uses them because they mostly suck. Google's DNS server serves something like a trillion requests a day. It might be hard to see the business value for google but it is clear. Google makes money when people use the internet. The DNS server exists to make the internet the best it can be. If you have a shitty dns server you can always use googles. That means google is still making money off you. They won't kill something so integral o their basic monetization strategy.
  - drewbug01 6y ago
    I literally said: > They'll keep Google DNS running as long as it provides business value to them, and that's it. As in yes: I know it provides business value to them. > Use your brain Please be civil in your comments and refrain from insults. That’s not a charitable way to interact with others.
- jsnell 6y ago
  > I understand the point you're trying to make here The point was that it's a multi-dimensional space, and the OP should actually consider the product rather than automatically go all "lol, Google product, bet it gets killed". So if you think that "some old products were discontinued" is any kind of rebuttal, I clearly didn't make my point well enough. Of course that happens! The alternative is that any sufficiently old product would automatically become immortal, which would be a ludicrous idea. 8.8.8.8 clearly has a ton of users, which already differentiates it from basically everything on your list. It's also isn't something you could just put in a maintenance mode and forget about, unlike a lot of the things on your list, both due to the scale and due to the impact if it were to stop working. Something like "Google Showtimes" would not have been a big drain on resources for most of its lifespan... When there's a measurable cost to keeping a service running, the longevity does actually signal something about the business value. Products with no users get killed by all companies. Products with hundreds of millions of users don't get killed with one exception: to migrate the users to a different product for the same task. And that can't really happen with DNS! They can't replace the clients, nor force the clients to upgrade, and they can't change the protocol in a way that would force some kind of a migration. Even if they end up deciding that the service needs a full rewrite, the external interface will have to stay the same. And once you think about the specifics, it actually becomes kind of an interesting discussion to have! What are the circumstances that could lead to this service being discontinued? A complete migration from IPv4 to IPv6 might do it: part of the value of both this and 1.1.1.1 is that these are IP addresses that people can actually remember. Their IPv6 addresses do not have that property. Not holding my breath on that one though :-P And even if that migration ever finishes, it's plausible that operating systems start including a dropdown of well-known public DNS servers as one of the configuration options. Could they replace DNS entirely? Come up with a "QuikDNS" that starts off as proprietary, is implemented in only Chrome and Android but never replaces DNS outside of their ecosystem? Or instead of a proprietary protocol just stop supporting classic DNS and only continue supporting DNS over HTTP? I don't see the former, there's just not enough wrong with the standard protocols for that to be worth it. I could definitely imagine the latter happening at very long timescales (like, not for at least 10 years). At the point where 99% of the traffic is DNS over HTTP, maybe the cost benefit ratio stops being there for classic DNS. > and a lot of the things they killed were well used, long lived, and popular. I don't think the examples you posted really match that description. Most of them weren't ever popular, let alone when they were discontinued. Maybe Picasa was?
samoa42 6y ago
probably a win-win situation with google wanting to ensure users can use google services (foremost doubleclick) and users wanting to avoid crappy or censored isp resolvers. but gog not using the harvested data seems somewhat unbelivable

by jsnell 6y ago

willis936 6y ago

It’s not a joke. It’s a question of trust and authority. “Google DNS has been running for 10 years” does not negate Google’s management behavior or change the inherent risk of a free running for-profit entity being considered a DNS authority. You calling it a tired joke doesn’t change these things.

acdha 6y ago

On the business value front, I’d also bet that they see value from the data visibility front since that gives them a chance to identify malware and phishing domains early and the terms of service appear to allow aggregate activity data mining. It’s not as precise as Google Analytics, of course, but it works on many things which don’t have JS beacons and I’d be surprised if they couldn’t find a good deal of useful information from that stream.

drewbug01 6y ago

> Track record: 8888 has been running for more than ten years. I understand the point you're trying to make here - but long-lived products are still shut down by Google, fairly often. Here's a list of some Google products that have been / were around for 10 years or more before being killed (or are slated to be killed soon): - Chrome Apps - Cloud Print - Fusion Tables - Youtube Video Annotations - Google Search Appliance (17 years old!!) - Google Showtimes - Google Code - Picasa - Orkut - Postini That's not even close to the full list of things that were more than 10 years old when Google shut them down. I left a bunch off for brevity's sake - but browsing https://killedbygoogle.com https://killedbygoogle.com is a really eye-opening experience. They really have shut down a lot of stuff - and if you loosen the requirement to ">= 7 years" the list gets very, very long. Again - I understand what you're trying to say. But it's just not simply a "zero effort joke". Google has killed a lot of things, and a lot of the things they killed were well used, long lived, and popular. They'll keep Google DNS running as long as it provides business value to them, and that's it. (Edit: I'm bad at list formatting, sorry)

colinmhayes 6y ago
Use your brain. Google kills product because no one uses them because they mostly suck. Google's DNS server serves something like a trillion requests a day. It might be hard to see the business value for google but it is clear. Google makes money when people use the internet. The DNS server exists to make the internet the best it can be. If you have a shitty dns server you can always use googles. That means google is still making money off you. They won't kill something so integral o their basic monetization strategy.
- drewbug01 6y ago
  I literally said: > They'll keep Google DNS running as long as it provides business value to them, and that's it. As in yes: I know it provides business value to them. > Use your brain Please be civil in your comments and refrain from insults. That’s not a charitable way to interact with others.
jsnell 6y ago
> I understand the point you're trying to make here The point was that it's a multi-dimensional space, and the OP should actually consider the product rather than automatically go all "lol, Google product, bet it gets killed". So if you think that "some old products were discontinued" is any kind of rebuttal, I clearly didn't make my point well enough. Of course that happens! The alternative is that any sufficiently old product would automatically become immortal, which would be a ludicrous idea. 8.8.8.8 clearly has a ton of users, which already differentiates it from basically everything on your list. It's also isn't something you could just put in a maintenance mode and forget about, unlike a lot of the things on your list, both due to the scale and due to the impact if it were to stop working. Something like "Google Showtimes" would not have been a big drain on resources for most of its lifespan... When there's a measurable cost to keeping a service running, the longevity does actually signal something about the business value. Products with no users get killed by all companies. Products with hundreds of millions of users don't get killed with one exception: to migrate the users to a different product for the same task. And that can't really happen with DNS! They can't replace the clients, nor force the clients to upgrade, and they can't change the protocol in a way that would force some kind of a migration. Even if they end up deciding that the service needs a full rewrite, the external interface will have to stay the same. And once you think about the specifics, it actually becomes kind of an interesting discussion to have! What are the circumstances that could lead to this service being discontinued? A complete migration from IPv4 to IPv6 might do it: part of the value of both this and 1.1.1.1 is that these are IP addresses that people can actually remember. Their IPv6 addresses do not have that property. Not holding my breath on that one though :-P And even if that migration ever finishes, it's plausible that operating systems start including a dropdown of well-known public DNS servers as one of the configuration options. Could they replace DNS entirely? Come up with a "QuikDNS" that starts off as proprietary, is implemented in only Chrome and Android but never replaces DNS outside of their ecosystem? Or instead of a proprietary protocol just stop supporting classic DNS and only continue supporting DNS over HTTP? I don't see the former, there's just not enough wrong with the standard protocols for that to be worth it. I could definitely imagine the latter happening at very long timescales (like, not for at least 10 years). At the point where 99% of the traffic is DNS over HTTP, maybe the cost benefit ratio stops being there for classic DNS. > and a lot of the things they killed were well used, long lived, and popular. I don't think the examples you posted really match that description. Most of them weren't ever popular, let alone when they were discontinued. Maybe Picasa was?

samoa42 6y ago

probably a win-win situation with google wanting to ensure users can use google services (foremost doubleclick) and users wanting to avoid crappy or censored isp resolvers. but gog not using the harvested data seems somewhat unbelivable