Amazon customers consider on-premise servers after cloud service outage

Amazon cloud customers are considering deploying their own servers after a second Amazon Web Services (AWS) outage brought their websites down.

Amazon cloud customers are considering deploying their own servers after a second Amazon Web Services (AWS) outage within four months brought their websites down.

Amazon and Microsoft European cloud services were down at the weekend after a lightning strike caused power failures at their datacentres in Dublin.

The lightning strike took out the main power supply and affected part of the phase control system that synchronises the back-up generator plant, causing a disruption to service of Amazon's Elastic Compute Cloud (EC2) cloud computing platform for the second time this year, as well as affecting Microsoft's Business Productivity Online Suite (BPOS).

Paul Johnston, CEO of mobile apps agency Padajo, uses Amazon's Simple Storage Service (S3), EC2 and Relational Database Server (RDS) to run a mobile communication application for a client's sales team. The outage stopped the application working completely.

Padajo has made an emergency update but Johnston fears some data may have been lost.

"We have certainly lost data, hopefully only temporarily, that the client usually receives on a Monday morning," he said.

As a result, the company is now considering introducing its own database server in the future as the complexity of Amazon's service has made it more difficult to get back online.

"We chose the RDS because it was felt it was more stable and robust than running an EC2 instance with a database server, because we believed Amazon would have thought through their disaster planning better than we can - at least, that's what I thought," said Johnston.

"As it turns out, if we had run our own database server via an EC2 instance instead of RDS, we'd probably have been back up a lot quicker, and so I'm seriously considering doing that in future or at the very least, ensuring there is a constant backup on non-Amazon servers in future to account for just this eventuality. If it happens again, I'll leave Amazon AWS and find another cloud hosting company," he said.

"I'm frustrated because cloud was meant to be better than having to run your own systems and servers, but in this instance, it's caused more problems than it solves due to Amazon's frailties and not my own. "

Authority ranking website PeerIndex also suffered from the outage. In a statement, PeerIndex said the disruption to Amazon's EC2 datacentre in Ireland affected PeerIndex servers along with "many other services", resulting in the website being down.

"We are working with Amazon to restore PeerIndex as fast as possible," the statement said.

Game-selling platform IndieCity said its site was down for a total of 12 hours as a result of the Amazon outage.

Scott MacKintosh, community manager at IndieCity, said the outage damaged the start-up company's reputation: "As an online retailer, our website being online is a hugely important factor. The main damage was to independent developers trying to integrate and upload their games to our website. The Amazon outage meant they couldn't access the site or services."

Amazon's Elastic Cloud Compute service also suffered an outage earlier in the year.

Check the status of Amazon's services here.

Read more on IT risk management