Hacker News Reader

Comment by ryandvm | Hacker News Reader

Man, I did not have "AWS us-east-1 will only have TWO 9s this year" on my bingo card.

For those of us who have been using AWS for almost 20 years now, I can't imagine why anyone would willingly choose us-east-1 for anything. It is the oldest, highest traffic, most critical path region and is subject to turbulence.

It can make sense to depend on the thing that will attract massive worldwide attention if/when it goes down. Or, more likely, it's just a default people don't change.

By some logic, that would mean it is the most battle-tested and highest-stakes (and therefore most carefully-managed) choice. I.e. reasons in favor.

Not that I disagree with you, but maybe not for the reasons you say (:

> By some logic, that would mean it is the most battle-tested and highest-stakes (and therefore most carefully-managed) choice

As someone who used to work on the inside, us-east-1 has the biggest pile of legacy workarounds for internal AWS issues, it has a variety of legacy API behaviours that don't exist in other regions, and because everyone picks it as the default, it has significantly more pressure on contested resources (i.e. things like spot instance pools).

Plus since it's the default in all the tooling, if you ever decide to go multi-region, you'll find tons of things break right away.

I think it is a little complicated. For example, your service might be using full failover but you use API from other service which are down.

Or you might use BART to come to work and you got stuck: https://www.kqed.org/news/12060687/bart-resumes-service-but-...

ha! I saw another comment on here talking about how ec2 doesn't need to be held to the same standard as the power company because it's not as important as real infrastructure.

wish I'd already had this link in my back pocket. our industry needs to take its job, as a whole, much more seriously.

“Global” and “edge” services such as IAM, Route53, CloudFront and so on have dependencies on us-east-1, so even if you don’t think you do, you probably do.

Well, we didn't, but some of our third party softwares did. Hard to avoid.

Wait, was the whole region affected? Like even if you had an EC2 instance?

No, we run on US East 1 but only EC2. Everything was running smoothly!

Our strategy has always been to use as little higher abstractions from cloud providers as possible. Glad we went this way, saved us quite a bunch of SLA breaches today! I am confident to say that it's "best of both worlds". We get great availability zone redundancy by AWS without having to rely on and pay for all those PaaS stuff the cloud giants offer. Also, we can "fairly easy" migrate to any other cloud provider because we only need Debian instances running.

Yes, it was. We have EC2 instances that we turn on as-needed, and at times were unable to start said instances.

Reply on news.ycombinator.com