Re:AWS EC2/RDS Outage in us-east-1
by deusex_ on Saturday August 31, 2019 @03:54PM
(#20848449)
Your us-east-1f is not the same one as on other accounts. The letter is randomly assigned to the AZ to spread load.
[ Parent
| on HN ]
Re:AWS EC2/RDS Outage in us-east-1
by whalesalad on Saturday August 31, 2019 @04:32PM
(#20848639)
Interesting, never knew that. I guess that is why the announcements never explicitly pointed to a single AZ by name.
Essentially everyone picks "a" because it's the first az. There's some internal mapping to "your az a is actually datacenter q". You can kind of figure out which AZs match across accounts if you've got enough accounts you can send traffic between.
I've been told that the "a" AZ you get was the least populated at object creation time (ie the first time you make an object that lives in an az), but I don't know how valid that is.
Re:AWS EC2/RDS Outage in us-east-1
by whalesalad on Saturday August 31, 2019 @05:40PM
(#20848938)
Aside from spinning up an EC2 node in each AZ and doing ping or tracing tests I wonder if there is a quick-n-dirty way to map AZ’s between different AWS accounts. I’ve never had to approach that scenario (cross account, low latency requirements) but in the future I’ll keep this in mind.