The innovative solution of datacenters, and their continuous evolution over time has allowed the development of data processing centers for various purposes (power, consumption, data, virtualization…), to address the needs of customers more carefully and saving unnecessary resources. In the same way, certain ranges of datacenters have emerged depending on the roughness of operation and cost that consumers are willing to invest.
In these ranges we differentiate the TIER classification, in English, which is a standardized methodology that defines and measures the availability time of a datacenters, as well as other aspects to consider such as the redundancy of connections and devices. They are useful in measuring the performance of a datacenters, its investment as well as its return.
Also, 4 levels or Tiers exist, each with its characteristics:
First, Tier 1, considered the most basic level in terms of performance and availability, which is used by small businesses due to its low cost. This offers components without redundant capacity, as it would be at the level of UPS connections, such as IPS. They are also characterized by guaranteeing 99.671% availability, or in other words, no more than 28.8 hours of bankruptcy per year.
Then there is Tier 2, already as an intermediate Tier, which is affected by a higher cost due to its better performance. However, to date, a datacenters is not considered reliable if it is not in Tier 3 or 4. Thus, it encompasses everything in Tier 1, plus devices with redundant components, including the most critical, system of redundant refrigeration and energy. Also, it guarantees an availability of 99.741%, or what comes to be the same, no more than 22 hours of annual bankruptcy.
It would then move on to Level 3 Tier, which offers a big leap over its previous version. This offers a great improvement in availability, thus guaranteeing 99.982% of operating time, or a maximum of 1.6 hours of annual bankruptcy. It brings together all the benefits of Tier 1 and Tier 2, as well as offering dual power supplies, as well as multiple output links, fault tolerance and 72 hours of protection against blackouts.
Finally, we come to Tier 4, the best of the 4 Tiers, and the benchmark for availability and redundancy. This at the Spanish level, is usually found in companies with considerable turnover (BBVA, Sabadell, Telefonica). At the feature level, it groups the benefits of Tier 1, Tier 2, and Tier 3, as well as ensuring that all components are completely fault-tolerant, including data links, storage, air conditioning, or electrical power, among others. In terms of availability, it guarantees an incredible 99.995%, or in other words, no more than 26.3 minutes per year of service interruption. It is completely redundant, thus offering 0 SPOF (Single points of failure).
Finally, it is important to consider that these 4 levels of datacenters exist, in order to be able to face the choice of the same considering the availabilities of each one, as well as the redundancies and benefits that they offer, also considering the cost of each one.
Jaume Campeny and Alba Massa