Not Failsafe: Planning For Cloud Outages4:00 PM EST Fri. Nov. 16, 2012
Disaster recovery for cloud services gained attention following the outage at Amazon Web Services' data center in Virginia in October.
The Oct. 22 outage, which lasted most of the day, occurred when AWS' Northern Virginia data center complex went down. Earlier this year, a power outage cut services to customers in the same Northern Virginia data complex for about six hours. Other high-profile outages this year affected customers of Microsoft Windows Azure, GoDaddy and Google.
To prepare for cloud outages, solution providers recommend focusing on fundamentals -- ensuring geographic diversity of data centers, redundant contracts and remote backup whenever possible.
Kevin Chu, director of systems and infrastructure with digital marketing at technology company Digitaria, said the Amazon outages did not appreciably affect his company's clients because Digitaria had service contracts for them with Amazon in two service areas, as Amazon recommends.
He said geographic diversity of data centers is key for disaster recovery, something all solution providers should recommend to their clients, even if it adds costs.
Jeremy Pryzgode, CEO of Stratalux, a Platform-as-a-Service provider that builds and manages Amazon cloud solutions for customers, has contracts with Amazon in diverse regions.
He said he's also considered entering service availability contracts with two cloud providers for one client.
"In an ideal world, with two contracts you would switch to another provider, like Rackspace," he said. "But these kinds of switches are really expensive. You need to move your services into two different cloud providers which, depending on the APIs, may be more expensive."
Paul Clifford, president of Davenport Group, a solution provider and Dell partner, also recommends geographic diversity of backup with cloud providers. But he said it's up to the solution provider to make sure the backup plan works, even in the event of a failure of a cloud hosting provider.
Clifford recommends customizing a disaster recovery plan for every client. "Every customer is different," he said.
"We have too many examples of these [cloud hosting centers] going down," Clifford said. "Every CIO knows nobody is immune from systems going down.
"We have to build environments that ensure that no matter what happens I can rebuild it. If it's Dell or any other service provider, can they ensure I can come back and get the data, and then can they prove it to me," he added
PUBLISHED NOV. 19, 2012