3 ms·
Yeah, vibes with my experience working there. Toxicity beyond pale. Good riddens, they contact me every week to try to get me back. Fuck off Amazon, you have no
by goldenkey 5y ago
Yeah, vibes with my experience working there. Toxicity beyond pale. Good riddens, they contact me every week to try to get me back. Fuck off Amazon, you have no integrity, you abuse your employees from white collar to blue collar. You literally fired people who emailed Jeff Bezos to report abuses, after you told them to, when the NY Times "crying at desk" expose came out. Bezos instilled this shite culture into Amazon. And his midlet pawns carried out his half-baked decrees like those fucking wet noodle "Leadership Principles" which make no actual sense and aren't principles at all in any sense of the word. Brian Cantrill (of 0x1de computers) gives a great breakdown of how sacrosanct Amazon's leadership principles are: https://youtu.be/9QMGAtxUlAc https://youtu.be/9QMGAtxUlAc
Principles are things like Integrity, Honesty, Respect -- not Amazon's fuckwit bulletpoints like "Be right a lot" or "Frugality."
Technologically, AWS is built like a fucking house of cards with spit, tape, and glue. There is no real cohesion or organization, just a bunch of teams doing their own practices and gluing their rando codebase into the Yellow Console. Sorry, but just making the frontend look unified doesn't undo what the backend spaghetti is... That's why the outages of late will continue. It's all bubbling up from the absolute toxicity in lack of basic humanism at the core.
- tobinfekkes 5y agoI logged in simply to comment a huge "thumbs-up" to any Brian Cantrill reference. I've got an entire playlist of all his various talks around the interwebs, just to listen (and re-listen) in the background while coding. It checks off all the education, tech, history and, critically, humor boxes.
- tandymodel100 5y ago> Technologically, AWS is built like a fucking house of cards with spit, tape, and glue. There Still making more money than GCP and Azure (both of which are worse to use, IMHO) so...who cares?
- Throwawayaerlei 5y agoWhen you're running a world wide service with lots of moving and connected parts this dramatically increases the danger you'll have catastrophes that will cost much or most of that market share. As a thought experiment, suppose all of AWS was down for a full week? What would happen after that? More realistically, how many times can us-east-chaos-monkey bring down various global services before a lot of its users start making moves to mitigate or eliminate their exposure to AWS? For a more concrete example, I'm working on a project that has an easy but critical use of cloud services. The more I read about Amazon following Microsoft into Ballmer era stack ranking madness, and the most AWS fails objectively fails, the lower the priority it gets for which cloud platform to support first and best.