ANTenna Blog -- How-To

A premiere example of cloud service—Amazon's Simple Storage Service (S3)—was down for eight hours on Sunday, providing another example of why that the cloud cannot do away with the need for servers inside the firewall.

The Service Health Dashboard for S3 reported the problem at 9:05 a.m. Pacific time Sunday (July 20) with the restoration of full service confirmed at 5:12 p.m., slightly more than eight hours later. Amazon blamed the problem on glitches in the "gossip protocol" by which the different system components exchange information about their states. The system had to be taken off-line to restart the gossiping process.

The blogosphere was rife with complaints from small businesses who relied on the service, and whose day was ruined. While customers normally have to apply for billing credit under their SLAs, Amazon said they would get credited automatically for this event.

The system also had an outage in February, and Amazon's service status page indicates that there is no shortage of drama behind the scenes, with a list of maintenance issues and glitches that you'd expect from any large system.

The cloud has undeniable utility, but inevitably it provides an additional point of possible failure to any enterprises that uses it.

In the final analysis, an enterprise's relation to the cloud is similar to its relation to the power company—when there is a power outage, all you can do is hope for an early resolution. In the meantime, it's good to have backup batteries. With the cloud, it's good to have in-house servers.

Go to the bMighty Server How-To Center