Everyone knew what the costs were going in.
AWS, cloud was NEVER the cheaper solution. It was never faster, it was never better.
It was easier, and easier came with a price. Like lambs to the slaughter lots of people embraced it. Amazon is a profitable company because of it.
[1]: https://www.agriculturedive.com/news/koch-foods-antitrust-ch...
[2]: https://www.computerweekly.com/news/366566360/Googles-data-e...
I've not thought about the unintended consequences of this, but it feels like a reasonable regulation to have.
The asymmetry is obviously meant to trap customers, which is anti-competitive.
Additionally, settlement free peering is usually based on having a roughly balanced traffic ratio (something like 2:1 or 3:1 counts as roughly balanced), attracting more ingress traffic by making it free vs having paid egress helps the provider balance their ratios so they may help their case for settlement free peering.
I kind of doubt it's easier. If you use your own servers you need some stuff to manage them. If you use cloud, you need some cloud engineers to manage your cloud infrastructure. On top of that, if your developers use cloud APIs and frameworks, they have to learn that.
I think the bread and butter for the big clouds however are customers like the shop I'm at (see sibling comment), we can do development work and put it on our clients (mostly non-IT that has no interesting in hiring their own department) and if we for some reason need to part ways it's easy to "hand-over-the-keys" since subscriptions can be handed over between organizations.
But getting customers back is usually equally easy since there was never any fuzz when leaving so when they've become dissatisfied with the "cheap" options they don't remember leaving being a problem, regardless the winner is the cloud provider (but the costs are still small enough for each client that nobody is unhappy).
1: We have a dozen or so regular clients and being in a relatively high salary cost country, while our cloud bill isn't insignificant it's still far less than one full time employee, even less so than having 24/7 on-call handled (we have large enough 24/7 critical clients).
2: The cloud provides broken down billing, it's easy to pass it to the customer and just point to it, having our own employee we would need to structure up the billing process somehow. Our customers are outside of the IT field and doesn't care as long as it works and doesn't cost too much (it's a mostly fixed expenditure and easily separable from our development hours).
3: We have one client that does on-prem (partly out of habit) that's actually looking at parts to the cloud, partly because they've expanded internationally but I wouldn't be surprised if it's a bit of testing the waters since they have to spend a fair bit of their time on security and stability (security doesn't go away entirely of course but I'd be more sure of my backup processes on a large cloud compared to just locally).
4: I'm a tad on the fence of cloud-api's:
On the positive side: No need to implement a bunch of systems(badly?) and no need to build/install monitoring tools for those (and the extra administration burden that comes with updating those tools for X number of clients).
On the negative: There's a tad of churn on some cloud-api's, luckily Azure is mostly stable with plenty of migration time until you come around to a customer again (reading about Goog's churn feels like horror stories), and yes there is a bit of vendor lock-in if you go all-out (luckily we've not seen any reasons to change over).
I used to provide services like that, on call, and the number of ops related interventions for my cloud customers were consistently higher than for those on colo setups.
It's like virtual lego - the bricks fit the same way as physical lego but the difference is you can rent as many virtual bricks as you like, almost instantly with little up-front cost.
You still need to know how to build the lego pirate ship either way.
I disagree. The company that put their infrastructure on AWS may well have made an informed choice. But the third party service hosted in AWS doesn’t have a choice, because AWS would punish their mutual customer with egress fees if the service moved out of AWS.
I don't think making an informed decision has anything to do with competitiveness in any case. Example I might make an informed decision to publish my app on the app store, but this doesn't mean its practices are not anti-competition.
Cloud wasn't predatory in this... the price has always been there. If you lease a car and they mandate you get your oil changed with them for a price, and the price sucks, that's YOUR fault.
No one got locked into the cloud and THEN the egress prices went up... everyone went in knowing this.
The industry needs to man up and own shitty decisions rather than double down on them forever.
I'd argue it's still quite predatory. Basically, they reel in users with other services/products and then charge extortionate prices for egress specifically with massive margins just because they can.
In certain use cases it becomes all or nothing, either you host everything on the cloud or nothing at all which is certainly abusive and highly anti-competitive.
No one hid the prices from any one... it was all very up front and out in the open. No more data centers no more systems admins, no more capacity planing just scale on a dime... Here is the price chart....
And everyone didn't want to do that HARD work any more. They could just grow and grow and grow and it would save them money.
You know what happens when your not gonna run out of storage or bandwidth or servers... Everything gets bloated.
SO sure, we can say abusive... Amazon abused all the poor stupid vp's who took the lazy way out and let their systems get so fat and bloated on button mashing "l" shaped engineers. Crying about the lock in, about the pricing after you signed up for the service is your own fault. Take the hit and move on!
Knowing the cost beforehand and anti-competitive are two different things.