Amazon cloud outage makes the Internet suck

Major web sites such as Reddit are currently experiencing outages today due to issues at a Northern Virgina-based cloud server operation hosted by Amazon. According to a message from Amazon's Service Health Dashboard:

We are currently experiencing degraded performance for EBS volumes in a single Availability Zone in the US-EAST-1 Region. New launches for EBS backed instances are failing and instances using affected EBS volumes will experience degraded performance.

The outage started around 1:38 pm Eastern Time and Reddit has been bouncing off and on as of this writing, showing its visitors its rather friendly and cute message above. Venture Beat reports that other sites affected by the Amazon outage include Airbnb, Flipboard, Coursera and more.

This is not the first time that Amazon's cloud servers in Northern Virginia have had problems. In July, a massive electrical storm caused several of those servers to go offline for several hours. This caused services such as Netflix and Instagram, among others, to be inaccessible for that period of time. At the time, Amazon said a backup generator that failed to kick in was part of the reason for the outages. There's no word yet on what specifically is causing today's issues.

Source: Amazon | Image via Reddit

Neowin Live - This event has concluded

Report a problem with article
Previous Story

Hulu Plus app for Windows 8 released

Next Story

Western Digital reveals Windows 8-based portable hard drives

33 Comments

Commenting is disabled on this article.

I think people are missing the point of cloud.


Cloud is NOT meant to never go down. All technology will go down at some point. Even if these services are being hosted internally the same outage could still happen. The benefit is that you are not maintaining the infrastructure to run the services and you have an agreed to SLA. Hence the SLA…A Service Level Agreement that says; We will provide you this, for this cost and it will be available to you no less than this period of time. SLAs are in place to allow down time.

So how does a little outage make Amazon, Cloud and Internet suck? If Dropbox or other online storage goes offline for one day or maybe two days over a year…Do those services now SUCK. If you think it does, then go buy a 30K storage array, pay for a high bandwidth data link, create custom software that will sync all your devices and give you real-time access to your data EVERYWHERE at once. And then duplicate that solution in another GEO area and pay for that too, and finally...Guarantee yourself and/or your customers/users that it will never go offline, even for one day.

Cloud is great. DataCenter technology being provided to those who cant afford it and/or don't know how to do it, for a reasonable monthly cost and an internet connection. No internal IT staff required to operate it, etc….Having worked in I.T. for 15 years at fortune 500 companies....The same crap happens to them all the time and more than 2 days per year...And for home users with laptops and desktops and tablets and smartphones....Its a guarantee they are going down and losing data more than 2 days per year...

Cloud, Amazon AWS and others services like this are the best damn thing to come to the internet in a while...

this is what....the 3rd of 4th outtage in the last 2-3 years from Amazon? I'll say it again, the cloud is a dumb idea. It's great for things like Dropbox, photosharing, etc...but for mission critical websites/phone systems/CRM's, etc, it simply won't work.

We use Salesforce, and Contactual. if one of contactual's datacenters go down, so does our inbound lines. Last time 2 contactual servers went down, our customers got a busy tone for over an hour. That is unacceptable, or at least would be if I ran my own business.

The Cloud has it's place, but not for mission critical stuff. You'll see these stories again.

I wonder if Reddit and other sites that are only using one availability zone have plans to take advantage of multiple zones if their budget allows it.

WinRT said,
guess is time to move to Azure

Sheesh guy. Azure is worst in everything, shady billing process, expensive, unstable, slow and without counting the usual "microsoft surprises".

Brony said,

Sheesh guy. Azure is worst in everything, shady billing process, expensive, unstable, slow and without counting the usual "microsoft surprises".


what the hell? no

the420kid said,
i read a great article today about major cloud outages. putting your whole biz in the cloud is For Dummies :-P

Tell that to Apple whose services are entirely run from Amazon and Azure.

deadonthefloor said,

Tell that to Apple whose services are entirely run from Amazon and Azure.

The only thing i'd use the "Cloud" for is CDN, obviously me placing my content over multiples places is going to be more costly than the "Cloud" just to host some data

thealexweb said,

No one has 100% uptime, Azure goes down occasionally too.

I've always found Azure to have much better failsafes - as a rule they generally try to keep multiple hosted copies of your service located in different data centres around the world, so you should be fine in if one data centre goes down. When an Amazon data centre goes down... their fail safes don't seem to function quite as well. Which is surprising, given how many times it's gone down in recent times.

... of course, neither of them are perfect though.

Edited by ~Johnny, Oct 22 2012, 8:09pm :

alwaysonacoffebreak said,
And that is exactly why you go with full physical servers not cloud. Serves you right, reddit.

Oh ok

alwaysonacoffebreak said,
And that is exactly why you go with full physical servers not cloud. Serves you right, reddit.

And you expect them to pay for and manage their own redundant datacenters?
The economies of scale are what make the cloud appealing.

deadonthefloor said,

And you expect them to pay for and manage their own redundant datacenters?
The economies of scale are what make the cloud appealing.

That is the saddest excuse ever for a site as big as reddit. If I can co-locate (behind L3) my servers they can. Backed up daily without problems.

Edited by alwaysonacoffebreak, Oct 22 2012, 8:33pm :

alwaysonacoffebreak said,

Ever heard of co-location? You take your server, put it in a server farm behind L3 where you got 24/7 access. Hell if I can do it they can, I don't earn that kind of money in 10 years what they grab in 1.

Also all my content is backed up to another datacenter daily. I don't see where exactly is the problem? They are just trying to be cheap. That's it.

This isn't as simple as "if I can do it, they can". The cost of scaling to the size of reddit is not linear. I would not be surprised if it would cost reddit twice as much to host their own servers.

alwaysonacoffebreak said,
And that is exactly why you go with full physical servers not cloud. Serves you right, reddit.

Not necessarily, but they could have used them for redundancy for when Amazon Cloud goes down. That way they could have cut over to them.

alwaysonacoffebreak said,
And that is exactly why you go with full physical servers not cloud. Serves you right, reddit.

Reading the comments here, it's clear that a many people simply don't understand what the cloud is and what it offers.

The key point above is that only one out of many Amazon DC's was affected. If websites as big as Reddit are stupid enough to only make use of one availability zone, then downtime is to be expected. This doesn't mean using multiple availability zones will eliminate downtime altogether, but it will definitely improve up time.