One feature of a data center that is considered very important is its uptime. A data center stores data that have to be accessible all the time. It also hosts all physical components, including processor, memory and other important computing components, which ensure that the data it stores can be accessed and modified whenever necessary. If the data center is right in the middle of a network system with very busy traffic, a very brief downtime can surely be a disastrous disruption. The uptime of a data center must be defined. The four tiers of data center are generally used to define it. Those four tiers are also usable to calculate various business aspects of a data center, including its performance, investment and return of investment.
Each of the four tiers of a data center defines how available it is, with tier four defining that a data center is almost 100% available and prone to failures. In brief, the four tiers of a data center can be described as follows.
TIER ONE provides the most basic infrastructure to a data center. It consists primarily of a single path for power and cooling distribution. No redundant component is available in this tier. This data center tier is guaranteed to offer 99.671% availability, meaning that in a year, a tier one data center is expected to be down for only 28.8 hours.
TIER TWO is quite the same with tier one in that it also consists of a single path for power and cooling distribution. However, unlike tier one, tier two also consists of redundant components, which allow it to offer 99.741% availability or 22-hour yearly downtime.
TIER THREE multiplies the paths for power and cooling distribution; however, only one path is active at a time. Redundant components are also available in this tier. Generally, tier three data center offers 99.982% uptime or 1.6 hours of yearly downtime.
TIER FOUR consists of multiple power and cooling distribution paths that are active at the same time and redundant components. It is a state-of-the-art data center that offers 99.995% minimum uptime and 0.4 hours of yearly downtime.