undefined | Better HN

0 points__alexs5mo ago0 comments

Is there any data on which AWS regions are most reliable? I feel like every time I hear about an AWS outage it's in us-east-1.

0 comments

theherk5mo ago

Trouble is one can't fully escape us-east-1. Many services are centralized there like: S3, Organizations, Route53, Cloudfront, etc. It is THE main region, hence suffering the most outages, and more importantly, the most troubling outages.

dangoodmanUT5mo ago

This is not true. There are a FEW high-level ones that have their dashboards there, but there's a reason you log into region-specific dashboards

Macha5mo ago

We're mostly deployed on eu-west-1 but still seeing weird STS and IAM failures, likely due to internal AWS dependencies.

Also we use Docker Hub, NPM and a bunch of other services that are hosted by their vendors on us-east-1 so even non AWS customers often can't avoid the blast radius of us-east-1 (though the NPM issue mostly affects devs updating/adding dependencies, our CI builds use our internal mirror)

donavanm5mo ago

FYI: 1. AWS IAM mutations all go through us-east-1 before being replicated to other public/commercial regions. Read/List operations should use local regional stacks. I expect you'll see a concept of "home region" give you flexibility on the write path in the future. 2. STS has both global and regional endpoints. Make sure you're setup to use regional endpoints in your clients https://docs.aws.amazon.com/IAM/latest/UserGuide/id_credenti...

bradhe5mo ago

us-east-1 was, probably still is, AWS' most massive deployment. Huge percentage of traffic goes through that region. Also, lots of services backhaul to that region, especially S3 and CloudFront. So even if your compute is in a different region (at Tower.dev we use eu-central-1 mostly), outages in us-east-1 can have some halo effect.

This outage seems really to be DynamoDB related, so the blast radius in services affected is going to be big. Seems they're still triaging.

gnaman5mo ago

your website loads for a second and then suddenly goes blank. There is one fatal errors from Framer in the console

fragmede5mo ago

It is dark. You are likely to be eaten by a grue.

bradhe5mo ago

Thanks...probably an issue with Framer during the outage.

Thev00d005mo ago

Anywhere other than us-east-1 in my experience is rock solid.

linsomniac5mo ago

Agreed, my company had been entirely on us-east-1 predating my joining ~12 years ago. ~7 years ago, after multiple us-east-1 outages, I moved us to us-west-2 and it has been a lot less bumpy since then.

Dave3of55mo ago

If you're using AWS then you are most likely using us-east-1 there is no escape. When big problems happen on us-east-1 it affect most of AWS services.

nodesocket5mo ago

I don't recommend to my clients they use us-east-1. It's the oldest and most prone to outages. I usually always recommend us-east-2 (Ohio) unless they require West Coast.

dr-smooth5mo ago

and if they need West Coast, it's us-west-2. I consider us-west-1 to be a failed region. They don't get some of the new instance types, you can't get three AZs for your VPCs, and they're more expensive than the other US regions.

j / k navigate · click thread line to collapse

0 comments

theherk5mo ago

dangoodmanUT5mo ago

This is not true. There are a FEW high-level ones that have their dashboards there, but there's a reason you log into region-specific dashboards

Macha5mo ago

We're mostly deployed on eu-west-1 but still seeing weird STS and IAM failures, likely due to internal AWS dependencies.

donavanm5mo ago

bradhe5mo ago

This outage seems really to be DynamoDB related, so the blast radius in services affected is going to be big. Seems they're still triaging.

gnaman5mo ago

your website loads for a second and then suddenly goes blank. There is one fatal errors from Framer in the console

fragmede5mo ago

It is dark. You are likely to be eaten by a grue.

bradhe5mo ago

Thanks...probably an issue with Framer during the outage.

Thev00d005mo ago

Anywhere other than us-east-1 in my experience is rock solid.

linsomniac5mo ago

Dave3of55mo ago

If you're using AWS then you are most likely using us-east-1 there is no escape. When big problems happen on us-east-1 it affect most of AWS services.

nodesocket5mo ago

I don't recommend to my clients they use us-east-1. It's the oldest and most prone to outages. I usually always recommend us-east-2 (Ohio) unless they require West Coast.

dr-smooth5mo ago

j / k navigate · click thread line to collapse