Cloud Application Architectures

Service Levels for Cloud Applications

When a company offers a service—whether in the cloud or in a traditional data center—that company generally provides its customers with a service level agreement (SLA) that identifies key metrics (service levels) that the customer can reasonably expect from the service. The ability to understand and to fully trust the availability, reliability, and performance of the cloud is the key conceptual block for many technologists interested in moving into the cloud.

Availability

Availability describes how often a service can be used over a defined period of time. For example, if a website is accessible to the general public for 710 hours out of a 720-hour month, we say it has a 98.6% availability rating for that month.

Although 98.6% may sound good, an acceptable value depends heavily on the application being measured—or even what features of the application are available. If, for example, Google’s spider is down for 24 hours but you can still search and get results, would you consider Google to be down?

Most people consider a system to have high availability if it has a documented expectation of 99.99% to 99.999% availability. At 99.999% availability, the system can be inaccessible for at most five minutes and 15 seconds over the course of an entire year.

Luck Is Not High Availability

Referring to a system as a high-availability system describes what you can expect in terms of future uptime—not what you happened to get one year when you were lucky. The ...