Comment by f0e4c2f7
3 years ago
Wow I'm really sorry this happened to you. I do believe your conclusions are correct about AWS. Azure generally seems fine too.
Google's aversion to customer service makes them extremely dangerous as a cloud provider in my mind. This also goes for other critical business services, like Gsuite (I know! It's convenient).
If you have GCP or Gsuite now and you're trying to evaluate how big of a deal it is my suggestion would be to pick up a phone and attempt to talk with someone at Google about your account. This experience can act as a preview of what that process might look like when services are turned off.
If you try to call Amazon on the other hand it feels like Jeff Bezos might hop on the call if things aren't going well.
I've had some bumpy points with Azure, but if I really can't fix something I can pick up the phone and they'll try to help. The idea of having critical business services with a company that you can't actually talk to is terrifying.
Jeff is famous for sending “?” Emails to his s-team of direct reports for customer issues they mailed him about. Those ? emails derailed many a roadmap.
I know it’s just anecdotal, but once you have some spend (around 1k a month I’m guessing), you get an account executive who I’ve been able to reach on the phone whenever I’ve needed. In fact they tried talking to me on the phone more than I wanted.
Google engineers are not known for being super available though.
My rep was able to get me connected to a few tech leads on various cloud products and one time when we were hitting weird quota limits the team protectively reached out to me to ask what the heck we were doing. Maybe we were just lucky, but I’ve had similar experiences in AWS and Azure too.
One time I emailed jeff@ and then a few days later someone called to talk to me about my Kindle on Android bug. I got a workaround and it was fixed in the next version.
And I've talked to a support human at Google before, about our GSuite.
I’ve had phenomenally good support from AWS. I recently had a problem where a nat instance was opening more than 55k simultaneous connections (don’t ask) and AWS support helped me track down which ecs service it was coming from. I’ve had utter shit support from GCP (mostly around “hey can I get this API limit raised). My best support from GCP was from our customer evangelist. He actually knew stuff. His turnaround time was 3-5 days. I eventually just stopped filing support cases because they were useless.
Also currently actual AWS staff is answering on the new community support forum re:post, I’ve been taking full advantage of that.
There are some things which AWS suck at unless you have an enterprise support plan, like having a deep review of a technical issue in their products, and having visibility into product bugs in general.
Even then after reporting a significant issue like Aurora query execution being non-deterministic in a very specific case, I only found out it was solved months later after working around it from reading patch notes.