Hacker News Reader

Comment by zacharyvoase | Hacker News Reader

By what stretch of the imagination is this icon suitable for representing a total loss of availability due to a power outage?: http://status.aws.amazon.com/images/status2.gif

Is this not a 'service disruption' situation? At the bottom of the page, the yellow icon is associated with 'performance issues'.

If there's one thing that's shocked me about AWS, it's the total failure to acknowledge the severity of service disruptions. Like the above case, or the fact that a 3-hour loss of connectivity is displayed on the service history as a green tick with a small 'i' box: http://oi46.tinypic.com/x5qtch.jpg

View on HN

Or that there is absolutely no way to deep link to an ongoing outage, and users must reload, then expand the link every single time, or subscribe to an RSS feed.

AWS needs to blantantly copy Heroku's status system, which is worlds better for people needing fast updates on their infrastructure.

https://status.heroku.com/ vs http://status.aws.amazon.com/

Wow, that is so beautiful! I am in awe.

The whole Heroku site is phenomenal. I am constantly amazed every time i see it.

What's the deal with the % measuring? I know going below three nines is seen as "bad service", but three hours of outage in a month gives you 99.5%, not 99.97%

The 99.97% statistic has "may uptime" written next to it, so I guess it refers to the previous month.

They've reporting a power issue with a single US East availability zone. There are four EC2 availability zones in US East. Strange that they would cite orange as "performance issues", but it's certainly more appropriate than suggesting a complete service disruption.

It is absolutely affecting more than one availability zone. Just because people didn't lose boxes outside of one AZ doesn't mean there weren't service disruptions outside of it. I know multiple people, myself included, who STILL cannot launch new instances anywhere in US-East. The EC2 web console and API have also been down intermittently and throwing errors for hours.

If one availability zone is down, there will be suddenly be ton of extra people trying to start new instances in the other availability zones. Could there just be a shortage of available instances in the other availability zones? Perhaps Amazon don't keep enough spare capacity for this eventuality (nor would I expect them to)?

This certainly can happen. You can't rely on being able to launch enough replacement instances if a zone fails.

You can at least mitigate it by keeping utilization of your instances below what would be needed to handle load of one zone failing. I suspect to save money, many people are pushing utilization as high as possible.

This is what reserved instances are for. You can by low use reserved instances for disaster planning.

If an outage in one zone causes a brownout in the other zones, then the aren't independent "availability" zones in any logical sense. They may as well be on the same power supply.

They are independent IF you buy them in advance. If you wait till an actual outage then it's too late.

It is obvious we are dealing with the imagination of a marketing exec here. And that is a sick cynical place.

They do have a red icon if an event impacts an entire region. If a customer is correctly utilizing multiple availability zones, a failure in one zone should only impact the customer until they can fail over (Should be within minutes if they are automated).

Aren't multiple availability zones impacted though? I don't get how Heroku can be so widely impacted if its not for the fact that multiple Availabiltiy Zones have been impacted.

The power issue only struck 1 of the 5 zones in US East (Virgina)

I can't speak to why Heroku was impacted as much as it was. It could be that they have single points of failure, or run at a utilization level that makes losing a single zone difficult.

But the icon is meant to display AWS status, not the customer's. Friday's outage did warrant a red icon - unreachable instances, corrupted EBS volumes, failed API calls.

Clearly Amazon hired Baghdad Bob as their PR guy when he was looking for a new gig.

Apart from Apple's legendary secrecy, Amazon's EC2 is a solid #2 in terms of impenetrability.

This is not fair, Amazon publishes surprisingly detailed postmortem reports for AWS problems.

For embarrassing outages, sure, but have you ever had to try and find out why some of your instances have disappeared? Good luck with that. Gold support, which you'd think would be good, just prioritizes your emails, apparently little more.

Linode, in comparison, has always been straight-forward, personal, and as honest as a company in that space can be.

What's perhaps odd is Amazon's customer service for their material goods is usually superb.

It's a terrible choice, but seems inspired by the triangle shape used for warning signs.

Hey! At least Amazon.com is up!

An upwards triangle is used as a danger/warning traffic sign.

We had 0 downtime. The only thing that's screwed up is a read replica of a multi-AZ MySQL on RDS deployment. Amazon did not send any notification. Kinda annoying.

Reply on news.ycombinator.com