Hacker News Reader

Comment by jkingsman | Hacker News Reader

Because Amazon has $$$$$ in their SLOs, and it costs them through the nose every minute they're down in payments made to customers and fees refunded. I trust them and most companies not to be outright fraudulent (although I'm sure some are), but it's totally understandable they'd be reticent to push the "Downtime Alert/Cost Us a Ton of Money" button until they're sure something serious is happening.

View on HN

It literally is fraudulent though.

I don't think a region being down is something that you can be unsure about.

Oh, you can get pretty weaselly about what “down” means. If there is “just” an S3 issue, are all the various services which are still “available” but throwing an elevated number of errors because of their own internal dependency on S3 actually down or just “degraded?” You have to spin up the hair-splitting apparatus early in the incident to try to keep clear of the post-mortem party. :D

This is an incentive to dishonesty, leading to fraudulent payments and false advertising of uptime to potential customers.

Hopefully it results in a class action lawsuit for enough money that Amazon decides that an automated system is better than trying to supply human judgement.

Can someone just have a site ping all the GET endpoints on the AWS API? That is very far from "automating [their entire] system" but it's better than what they're doing.

Something like this? https://stop.lying.cloud/

Um, that's just a reskin of AWS's lying status page. No real new data there.

It should be costing them trust not to push it when they should though. A trustworthy company will err on the side of pushing it. AWS is a near-monopoly, so their unprofessional business practices have still yet to cost them.

> It should be costing them trust not to push it when they should though.

This is what Amazon, the startup, understood.

Step 1: Always make it right and make the customer happy, even if it hurts in $.

Step 2: If you find you're losing too much money over a particular issue, fix the issue.

Amazon, one of the world's largest companies, seems to have forgotten that the risk of not reporting accurately isn't money, but breaking the feedback chain. Once you start gaming metrics, no leaders know what's really important to work on internally, because no leaders know what the actual issues are. It's late Soviet Union in a nutshell. If everyone is gaming the system at all levels, then eventually the ability to objectively execute decreases, because effort is misallocated due to misunderstanding.

> It's late Soviet Union in a nutshell

How come an action of a private company in a capitalist country is like the Soviet Union?

Private companies are small centrally-planned economies within larger capitalist systems.

They talk about gaming the metrics. Something the Soviet Union was known for.

I can Google and see how many apps, games, or other services are down. So them not "pushing some buttons" to confirm it isn't fooling anyone.

Reply on news.ycombinator.com