Looks like I’ll finally have to replace that link in my resume after all.
It was useful to know when a copy of my printed resume was accessed online through the link I added on the footer, at least while the console for it was online.
Goo.gl has a namespace for 10 billion entries, it used to keep tracking/analytics data for each link, with a user interface, and it would happily generate them for stuff like Google Maps links.
How much money would you say it takes to even maintain a system like that, plus update its security, not to mention account for changing web standards, at that scale?
Probably like $50/month in cloud resources if you turned off all the extra stuff and only did redirects and kept it around in read only mode. You’d need to do some dev work up front and price that in as well, obviously.
Let’s take a conservative approach, and say there are:
only 1 billion links
each link only points to a URL of up to 100 characters in length on average (some will be 1000 or longer, but let’s hope some are 50 or shorter)
less than 10 billion daily hits total (that’s an average of 10/link)
the response time should be well under 50ms.
Now you’re looking at 100GB of raw data to put into a database, that needs to return 100K answers/second, in less than 50ms each, worldwide, 24/7.
What is your estimated cloud cost for something like 256GB of RAM, 128 cores, 10Gbps connect, replicated across several zones, and 1TB/day outgoing transfer?
That’s only for the redirect responses in read-only mode, nothing else. You will also need some maintenance to keep it 24/7, for when the server catches fire, or gets obsoleted, and when new exploits come up against your software stack.
The dumbest part is like, why? How much work is it really to keep goo.gl links around?
In 2018, Google wanted developers to move to Firebase Dynamic Links that detect the user’s platform and sends them to either the web or an app. Google ended up also shutting down that service for devs.
How much work is it really to keep goo.gl links around?
A lot.
Goo.gl has a namespace for 10 billion entries, it used to keep tracking/analytics data for each link, with a user interface, and it would happily generate them for links to internal stuff.
Just keeping it running would take some containers of server racks, plus updating the security, accounting for changing web standards, and so on.
Keep in mind this isn’t some self-hosted url shortener with less than a million entries and a peak of 10K users/second, that you can slap onto a random server and keep it going. It’s a multiple orders of magnitude larger beast, requiring a multi-server architecture just to keep the database, plus more of the same for the analytics, admin interface… and users will expect it to return a result in a fraction of a second, worldwide.
I think they’ve dropped the tracking already. Still, where’s the money in that?
They also can’t release the database, not without prior consent of the link creators, or risking exposing some login credentials some very smart people might’ve put in there.
I have probably saved hundreds of links from such a fate in my org. People there use them for everything even though the media they’re using them in allows them to be clicked (e.g. they’re not going out to print where someone has to type them in).
Thankfully, I’m in a position to un-shorten them before they get published. lol
It might be interesting to have a search engine or someone else who has built a massive list of links visible online generate unshortened forms now before Google shuts down the service.
Thanks. It says that there are already browser plugins that use their database, so looks like there’s already a way on both the scraper and user ends to programmatically avoid link rot here.
Signal is centralized, closed-source, not-selfhostable (edit: in any meaningful way) and requires being attached to a phone number. (Edit: server source is available, but self-hosting requires recompiling and distributing a custom app to all of your contacts to actually use it.)
Matrix is decentralized, federated, fully open source with multiple client and server implementations, self-hostable, and does not require being attached to a phone number.
Where the metadata goes I think is important as well.
All Signal metadata necessarily goes through Signal’s servers and is tied to your phone number, but not all Matrix metadata ever gets near the Matrix.org if you are using a different homeserver.
I think both are less than ideal in that regard, and I think Briar (strictly P2P) has a much better model for dealing with this at the expense of generally being a UX disaster.
That’s fair, though surely the metadata does go near matrix.org if either you contact people on matrix.org’s servers or use the fallback for calls since a lot of servers don’t have their own TURN server for that.
I thought Signal didn’t really have much metadata or is it more that it is extremely temporary to the case where it doesn’t matter?
And yeah, definitely agree that most extremely private (or at least marketed that way) messengers have a terrible UX and in a lot of cases UI too.
The server software appears to be available and updated now, which they’ve been spotty about in the past. I’ve updated to remove the closed-source part since that is not correct.
As for phone number: Signal still requires me to enter a phone number to create an account as of about 5 minutes ago.
Disclaimer: I don’t use any google services myself.
Because it is free, guaranteed to work as long as they keep it running and marketed well.
Plus since they were early into the game of tech online they have many services that all link together.
There aren’t many that will offer most users so much value for ‘free’.
Most alternatives will have some cost if you want as much space as google provides, either the same as google (user data) or monetary (which I semi agree with, hosting isn’t free and I’d rather pay money than with data). However, not everyone is in a position to pay with money and so data is usually what they pay with.
We use Google Forms and Sheets at work, precisely because easy for a bunch of us to access, and our boss is tight as fuck, so it being free is a massive draw.
I keep looking to other ways to perform the few functions we use, but ultimately I lack the knowledge and resources to roll my own.
Yeah, I lack the knowledge and reaources to roll my own too.
So I mostly rely on cryptpad for sharing/collaborating on different document types myself. I don’t think it is necessarily free for businesses, but I am unsure.
I’d assume that Google’s value – as with other link-shortening companies – came from being able to add information tracking whenever someone clicked on that link.
If you mean customer value, might be formats where people had limited space to include links like traditional Twitter (which was originally 140 characters in a post, whereas URLs have no specification-mandated character limit).
You are not logged in. However you can subscribe from another Fediverse account, for example Lemmy or Mastodon. To do this, paste the following into the search field of your instance: !technology@beehaw.org
A nice place to discuss rumors, happenings, innovations, and challenges in the technology sphere. We also welcome discussions on the intersections of technology and society. If it’s technological news or discussion of technology, it probably belongs here.
Remember the overriding ethos on Beehaw: Be(e) Nice. Each user you encounter here is a person, and should be treated with kindness (even if they’re wrong, or use a Linux distro you don’t like). Personal attacks will not be tolerated.
https://killedbygoogle.com/
Another one for the graveyard!
Looks like I’ll finally have to replace that link in my resume after all.
It was useful to know when a copy of my printed resume was accessed online through the link I added on the footer, at least while the console for it was online.
Could do similar things like adding a +resume to your email link
How much money they can save for this?
Probably with the saved money can’t even pay one single day of salary for the CEO
Oh, the CEO’s pay is secure from day one of the fiscal year. They’re trying to pay the cleaning staff with this.
What cleaning staff? Cleaning is just another part of your job now, this is purely for the shareholders.
Accurate 😬
URL shorteners in general, or just Google?
https://wiki.archiveteam.org/index.php/URLTeam
Goo.gl has a namespace for 10 billion entries, it used to keep tracking/analytics data for each link, with a user interface, and it would happily generate them for stuff like Google Maps links.
How much money would you say it takes to even maintain a system like that, plus update its security, not to mention account for changing web standards, at that scale?
Probably like $50/month in cloud resources if you turned off all the extra stuff and only did redirects and kept it around in read only mode. You’d need to do some dev work up front and price that in as well, obviously.
$50/month would barely scratch the surface.
Let’s take a conservative approach, and say there are:
Now you’re looking at 100GB of raw data to put into a database, that needs to return 100K answers/second, in less than 50ms each, worldwide, 24/7.
What is your estimated cloud cost for something like 256GB of RAM, 128 cores, 10Gbps connect, replicated across several zones, and 1TB/day outgoing transfer?
That’s only for the redirect responses in read-only mode, nothing else. You will also need some maintenance to keep it 24/7, for when the server catches fire, or gets obsoleted, and when new exploits come up against your software stack.
Does Google no longer want to pay for the Greenland .gl TLD?
The dumbest part is like, why? How much work is it really to keep goo.gl links around?
lmao
Yeah, shouldn’t be too hard to at least keep the existing links working in a read only state.
A lot.
Goo.gl has a namespace for 10 billion entries, it used to keep tracking/analytics data for each link, with a user interface, and it would happily generate them for links to internal stuff.
Just keeping it running would take some containers of server racks, plus updating the security, accounting for changing web standards, and so on.
Keep in mind this isn’t some self-hosted url shortener with less than a million entries and a peak of 10K users/second, that you can slap onto a random server and keep it going. It’s a multiple orders of magnitude larger beast, requiring a multi-server architecture just to keep the database, plus more of the same for the analytics, admin interface… and users will expect it to return a result in a fraction of a second, worldwide.
Good analysis, I agree and understand.
They could drop all the tracking though and only serve the public redirects. A much simpler product that would retain web links.
I think they’ve dropped the tracking already. Still, where’s the money in that?
They also can’t release the database, not without prior consent of the link creators, or risking exposing some login credentials some very smart people might’ve put in there.
Why does there have to be money in it when they’re sunsetting the service?
Google/Alphabet is a for-profit corporation, it makes no sense for them to do anything without some sort of profit.
That’s a whole lot of link rot about to happen.
The Jedis are going to feel this one
I have probably saved hundreds of links from such a fate in my org. People there use them for everything even though the media they’re using them in allows them to be clicked (e.g. they’re not going out to print where someone has to type them in).
Thankfully, I’m in a position to un-shorten them before they get published. lol
It might be interesting to have a search engine or someone else who has built a massive list of links visible online generate unshortened forms now before Google shuts down the service.
https://wiki.archiveteam.org/index.php/URLTeam
Thanks. It says that there are already browser plugins that use their database, so looks like there’s already a way on both the scraper and user ends to programmatically avoid link rot here.
GoogLOL
LOL¹⁰⁰
Googlel
- https://www.urbandictionary.com/define.php?term=Lel
Don’t build your online life around Google services.
Don’t rely on any company keeping a service running unless you’ve got a contract with them
Switch to Proton, Linux, Librewolf, Matrix, Gimp and Libreoffice.
I’ve been v curious abt matrix but it’s taken me years to get everyone I care abt on signal 😅
Signal is centralized,
closed-source, not-selfhostable (edit: in any meaningful way) and requires being attached to a phone number. (Edit: server source is available, but self-hosting requires recompiling and distributing a custom app to all of your contacts to actually use it.)Matrix is decentralized, federated, fully open source with multiple client and server implementations, self-hostable, and does not require being attached to a phone number.
Sure, some of those things are accurate (some are accurate-ish). However, there is way more metadata with Matrix than Signal.
To be clear, I use both, but Matrix’ metadata problem bothers me.
Where the metadata goes I think is important as well.
All Signal metadata necessarily goes through Signal’s servers and is tied to your phone number, but not all Matrix metadata ever gets near the Matrix.org if you are using a different homeserver.
I think both are less than ideal in that regard, and I think Briar (strictly P2P) has a much better model for dealing with this at the expense of generally being a UX disaster.
That’s fair, though surely the metadata does go near matrix.org if either you contact people on matrix.org’s servers or use the fallback for calls since a lot of servers don’t have their own TURN server for that.
I thought Signal didn’t really have much metadata or is it more that it is extremely temporary to the case where it doesn’t matter?
And yeah, definitely agree that most extremely private (or at least marketed that way) messengers have a terrible UX and in a lot of cases UI too.
Which part of signal is closed source? It no longer requires a phone number, you can use a username
I don’t think the server software is open source.
https://github.com/signalapp/Signal-Server
The server software appears to be available and updated now, which they’ve been spotty about in the past. I’ve updated to remove the closed-source part since that is not correct.
As for phone number: Signal still requires me to enter a phone number to create an account as of about 5 minutes ago.
Signal is open source, session copied their entire protocol for their own messaging stack. Client and server
Signal is fine for instant messaging.
Matrix is closer to Discord.
Why anyone uses a single Google product, I’ll never know.
Disclaimer: I don’t use any google services myself.
Because it is free, guaranteed to work as long as they keep it running and marketed well.
Plus since they were early into the game of tech online they have many services that all link together.
There aren’t many that will offer most users so much value for ‘free’.
Most alternatives will have some cost if you want as much space as google provides, either the same as google (user data) or monetary (which I semi agree with, hosting isn’t free and I’d rather pay money than with data). However, not everyone is in a position to pay with money and so data is usually what they pay with.
We use Google Forms and Sheets at work, precisely because easy for a bunch of us to access, and our boss is tight as fuck, so it being free is a massive draw.
I keep looking to other ways to perform the few functions we use, but ultimately I lack the knowledge and resources to roll my own.
Yeah, I lack the knowledge and reaources to roll my own too.
So I mostly rely on cryptpad for sharing/collaborating on different document types myself. I don’t think it is necessarily free for businesses, but I am unsure.
what were they used for? internal redirects like t.co? or something for customers? genuine question
I’d assume that Google’s value – as with other link-shortening companies – came from being able to add information tracking whenever someone clicked on that link.
If you mean customer value, might be formats where people had limited space to include links like traditional Twitter (which was originally 140 characters in a post, whereas URLs have no specification-mandated character limit).
Anyone could generate them, for free, and they came with analytics on the side. Google also generated them for sharing content from their services.
https://en.m.wikipedia.org/wiki/Google_URL_Shortener