Not all data centers are created equal. Today’s facilities come in all shapes and sizes, ranging from mobile micro-data centers that can be transported to provide remote storage and processing power to massive hyperscale facilities that house thousands of servers. Since these facilities are so versatile and perform so many different tasks, it can be difficult to categorize them based purely on function or size.
Developed by the Uptime Institute in the early 1990s, the Data Center Tier Classification system provides a helpful methodology for categorizing data centers on the basis of uptime reliability. After all, no matter what role a data center plays as part of a company’s IT infrastructure, its overall level of uptime reliability is perhaps the most important aspect of its performance. Each tier represents different levels of availability, hours of operation per year, and the infrastructure redundancy standards in place to preserve lifeline services.
Typically a modest on-premises data solution, tier 1 data centers offer very little in the way of infrastructure redundancy and experience high levels of system downtime. They have no backup systems in place to protect against power loss or cooling system failure and tend to lack the energy efficiency standards of higher tier facilities, which results in higher power and cooling costs.
A slightly more robust class of facility, tier 2 data centers are usually private data solutions that incorporate very basic backup systems to preserve system availability. They still tend to be inefficient in terms of power and cooling usage, either due to poor infrastructure design or an inability to manage processing workloads efficiently.
Considered the bare minimum standard for most enterprises, a tier 3 data center has redundant systems in place to limit their vulnerability to equipment failure and various operational errors. They utilize uninterrupted power supply (UPS) battery systems that continue to deliver power to servers while power is switched over to generator backup, helping them to maintain consistently high levels of uptime.
A tier 4 data center incorporates true fault tolerance into its infrastructure with full 2N redundancy. It has twice the resources and equipment necessary to maintain services even in the event of severe disruptions. Organizations that can’t afford to tolerate even a few moments of downtime generally turn to tier 4 data centers to meet their computing and colocation needs.
Uptime reliability refers, of course, to the percentage of time a data center keeps its customers’ services up and running over a set period of time (usually yearly or monthly). It represents the data center’s ability to handle disruptive events like a power outage, natural disaster, or data breach. A high level of uptime reliability indicates that a facility has the infrastructure redundancies and operations processes in place to manage unexpected problems and avoid common pitfalls (such as human error) that often lead to system downtime.
Small differences can have major consequences when it comes to uptime reliability. Every tier features uptime of at least 99%, with differences of less than one percent. This may lead an organization to think it’s safe to assume uptime is similar across tiers, especially if they also offer 100% SLA uptime guarantees. But these fractional slices of time can add up quickly. A tier 1 data center can be down 28.8 hours in the course of a year, while a tier 4 data center has an annual maximum downtime of 0.4 hours. Every second of downtime translates directly into lost revenue and missed opportunities, and those financial costs can become quite significant for organizations, to say nothing of the potential damage to their brand.
Despite these scary numbers, some businesses have a tolerance for downtime, especially planned downtime such as maintenance. Businesses that still rely primarily on batch processing can often easily restart after an outage, and there’s usually significant leeway in completion times.
For most data centers, their uptime reliability record is a byproduct of their redundancy standards. Tier 1 data centers, for instance, have a single path for power and cooling to the server equipment, with no redundant components. They typically do not have a backup cooling system or generator, so when they’re down—for maintenance, for instance—they’re completely down.
On the opposite end of the spectrum, a tier 4 data center is fully redundant. As 2N (or 2N+1) facilities, these data centers are outfitted with multiple cooling units, backup generators, power sources, and chillers. Even in the event of multiple equipment failures, they can continue to operate as normal. A typical tier 4 data center can sustain a power outage for as much as 96 hours, while a tier 1 data center will shut down completely once the power goes out.
While the tier classification system is a helpful means of categorizing data centers, it’s really just another way of assessing redundancy and uptime reliability. When evaluating a colocation provider, customers should carefully assess the facility’s uptime reliability record and infrastructure redundancy. Skilled operations teams and intelligent monitoring platforms (like vXchnge’s award-winning in\site) also play a major role in delivering high levels of system uptime. By selecting a facility with a long track record of system availability (preferably >99.99999%), organizations can ensure high levels of business continuity.