Slashdot Mirror
Major Outage At the Amazon Web Services
ralphart writes "The Northern Virginia datacenter for Amazon Web Services appears to be having a major outage that affects EC2 services. The Amazon Forums are full of reports of problems. Latest update from the status page: 2:49 AM PDT We are continuing to see connectivity errors impacting EC2 instances, increased latencies impacting EBS volumes in multiple availability zones in the US-EAST-1 region, and increased error rates affecting EBS CreateVolume API calls. We are also experiencing delayed launches for EBS backed EC2 instances in affected availability zones in the US-EAST-1 region. We continue to work towards resolution."
This will become quite the event in data warehouse circles I bet, because the cost of 'being in the cloud' just doubled; it's not enough to buy storage from one provider. The "always there" quality that's supposedly the benefit of cloud storage is a facade.
You can buy from one provider -- every major cloud provider has multiple availability zones. But yes, lots of people buy in only one zone because it's cheaper, and then suffer for that mistake -- in situations just like this.
Actually in the case of EC2 the smart thing would have been to have your instances spread over different availability zones...
"Back on topic, I just don't see all these day-long outages that apparenty seem to happen all the time in companies that haven't moved their servers to The Cloud(tm)."
You must not get out much. Atlantic.net I had a 11 hour outage due to the staff not understanding how to update a Cisco router. Then a 4 hour outage when they screwed up billing and shut down our service with no warning. Then there was that time they didn't like our DNS traffic and shut down DNS with no warning or notice. That was a fun hour or so of me trying to figure out why our applications were having issues.
So.. we go to replace them and one of the places we visit(that was highly recommended), first words outta their mouth was "So I'm sure you heard about that 6 hour outage we had xxx back. Here's what we've done to make sure it doesn't happen again..."
Long outages happen. I'm pretty much in the firm belief that if your app can't scale out across large geography automatically the only thing giving you solid uptime is pure luck. Any time someone wants 5 9's with a single data center I just laugh. But setting up your app and data to work that way takes Work.
Which, btw, would've also prevented outages for companies here. Only 1 zone was affected, EC2 also has zones on the west coast, Ireland and Asia pacific. If you built your app to use those and balanced via ELB, you likely wouldn't be impacted with this outage. But again, that takes $$$. Most companies don't want to spend that and frankly most companies probably don't need to.
his will become quite the event in data warehouse circles I bet, because the cost of 'being in the cloud' just doubled; it's not enough to buy storage from one provider. The "always there" quality that's supposedly the benefit of cloud storage is a facade.
The cloud doesn't have to be perfect - it just has to be as good in the eyes of VPs as the contractors they'd otherwise hire to run their internal datacenter. What's the value of an IT guy in the eyes of an MBA? Yeah, this sort of reality check wont phase them at all.
Socialism: a lie told by totalitarians and believed by fools.