Amazon outage an early lesson for cloud computing

Amazon is struggling to overcome an outage in its cloud computing services, prompting industry pundits to underscore the need for enterprises to back up critical services.

The outages began last week and have lingered on the company's Elastic Compute Cloud (EC2) and Elastic Block Storage (EBS) platforms for several days, causing temporary issues for a number of cloud services and web sites.

Attributed to a 'networking event' that slowed and limited service, the exact cause of the outage has not yet been disclosed. The company has said that it plans on investigating fully.

However, industry analysts are pointing to the outage as a black eye for Amazon, and a lesson for all enterprises considering moving to cloud platforms.

Enderle Group principal analyst Rob Enderle suggested that the incident should show enterprises that any essential cloud applications or services should be backed up with a contingency plan, similar to that of an on-premise service.

"If the service seems like it is going to be unreliable, you need a fallback," Enderle told V3.co.uk. "You need either a backup service, or you need resources in place you can fall back on."

The issue is particularly important to retailers, and other sectors that could suffer serious financial damage from the prolonged failure of a cloud platform.

"You can take a slowing down of the service, but a complete disconnect puts you out of business," Enderle said.

Other analysts noted that the incident could serve as an example for cloud computing clients, and any firm looking to modernise its datacentre operations.

Clabby Analytics president Joe Clabby told V3.co.uk that similar pitfalls could await enterprises that have virtualised multiple servers on a single hardware platform.

Much as a firm that relies on a single cloud computing platform could suffer an outage when the service goes down, Clabby explained that businesses which have consolidated datacentres onto a few machines could suffer should a failure occur.

"You used to set up separate failover servers, but that kind of goes away with virtualisation," explained Clabby. "It highlights the need that the resource pool is up and running at the same time."