How much downtime can your organization tolerate?
Not all data centers are created equal—nor should they be. In a digital age, businesses increasingly rely on third-party data centers to deliver high-availability environments that support their mission-critical workloads. Yet availability can mean something different to each organization as their needs, expectations and budgets vary.
Data center uptime is generally discussed in terms of 9s, which means increasing degrees of 99%—99.9%, 99.99% and so on. In some contexts—like your chances of winning the lottery or your flight taking off on time—99% provides great odds of success. However, in terms of data center reliability, 99% uptime during the 365 days in a year translates to significant downtime—to the tune of 3.65 days per year or 14.4 minutes every day.
Every minute of downtime can impact your revenue, productivity, customer satisfaction and reputation. According to one study, 88% of large enterprises reported a loss of more than $300,000 per hour when a server fails. This is a conservative number, as 40% of enterprises in an Information Technology Industry Council (ITIC) survey said the cost of an hour of downtime can range from $1 million to more than $5 million, excluding legal fees, fines and penalties. This number can climb to millions per minute if the outage interrupts a major business transaction or occurs during peak business hours.
Given the potential of a severe financial burden, businesses need a data center that can deliver the reliability that matches their uptime requirements. However, an Uptime Institute study found that 56% of organizations utilizing third-party data centers have experienced a moderate or serious IT service outage in the last three years that was caused by the provider. This looming risk makes having confidence in your data center’s capabilities essential.
Achieving a reliable, high-performing environment relies heavily on a data center provider’s ability to maintain the facility and its critical components, including uninterruptible power supplies (UPS), cooling units and generators, as well as to provide multiple power sources and distribution paths. In addition to building a highly available, resilient environment, data center providers must establish and integrate effective operational processes. This combination of physical systems and operational protocols is necessary to ensure uptime. The data center industry’s standard for uptime is five-9s—or 99.999%. Five-9s equates to 5.25 minutes of downtime per year—a far cry from the nearly four days of 99%.
The level of uptime an organization requires depends on its industry, the size and scope of the business, its customers’ expectations and the potential impact of downtime. Understanding your fault tolerance is important, as it can help you choose a data center that can deliver the right level of service without breaking the bank. While a data center with a 99.999% or higher uptime guarantee delivers a premium experience, not every organization needs this level of redundancy.
As an example, a small business without intense web traffic or 24/7 customer interactions may be able to take systems offline during non-business hours to perform routine maintenance. A global enterprise with heavy transactional demands and around-the-clock customer engagements does not have the same flexibility.
The good news is that there is a data center for every level of required protection and performance, each with its own price point.
The Tier Classification System offers a consistent and objective international standard for data center performance. Designed by the Uptime Institute in 2005, this system sets data center tier standards to create four distinct data center redundancy tiers: Tier I, Tier II, Tier III and Tier IV. Each tier is tightly defined based on a variety of factors, including data center redundancy levels, uptime guarantee, staff expertise, maintenance protocols and more.
The Uptime Institute tier levels are designed to provide insight into the minimum performance and availability expected of data centers within a tier. These standards are widely used by data center providers to self-evaluate and define their capabilities around uptime and redundancy. The capabilities of a data center escalate as you graduate from Tier I, which offers the least reliability, to Tier IV, with the most. Each tier includes the requirements of the previous level as you move up the tier ladder. Understanding the capabilities at each level can help you align your needs with an appropriately equipped data center.
As the most basic level, a Tier 1 data center extends no guarantee of redundancy for any critical systems. At a minimum, the Uptime Institute requires it to offer a UPS; a designated space for IT systems; dedicated cooling equipment that runs outside of office hours; and an engine generator. This infrastructure utilizes a single distribution path for the environment and only provides the power and cooling capacity to support the data center at full IT load. This lack of redundancy leaves the data center vulnerable to any kind of planned or unplanned disruption, including routine maintenance.
With limited resiliency, a Tier I data center guarantees a minimum of 99.671% uptime per year, which equals a maximum of 28.8 hours of downtime yearly.
Tier I data center list:
The Tier I customer. The simplicity of this tier makes it the least expensive option. Small businesses and start-ups without complex requirements or around-the-clock operations may be able to tolerate downtime during non-business hours for scheduled maintenance.
Tier II data centers build on the requirements of Tier I to include some redundant components. While this partial redundancy improves reliability to 99.741% uptime yearly—or 22 hours of downtime—these facilities still utilize a single distribution path for power and cooling and are still susceptible to unexpected interruptions.
Tier II data center list:
The Tier II customer. This bump in reliability makes Tier II an option for small- to medium-sized businesses (SMB) that need some guarantee of operational integrity but are still cost-conscious.
A Tier III data center offers additional reliability over Tier II in the form of N+1 redundancy and multiple power and cooling distribution paths. N+1 redundancy means the architecture offers the capacity to support the full IT load (N), and also offers an additional component (+1) for backup purposes, so performance is not impacted if a single component fails. Data centers in Tier III also utilize multiple distribution paths to ensure there are no interruptions if one path is unavailable. This level of redundancy offers concurrent maintainability, meaning that each critical component or distribution path can be shut down for planned maintenance without affecting the IT environment.
This tier significantly improves availability over the Tier II data center with 99.982% uptime for no more than 1.6 hours of downtime per year.
While Tier III is more robust than Tier I and Tier II, it is not completely fault tolerant. Tier III data centers can conduct routine maintenance without impacting service, but are still at risk of downtime during unscheduled events. Additionally, the risk of an outage increases during a maintenance window because the backup component is already deployed.
The Tier III customer. Large and growing companies may utilize Tier III data centers for their enhanced reliability and the efficiency they provide over the more complex protections of Tier IV.
The Tier III data center list:
Tier IV data centers are the most sophisticated tier certified by the Uptime Institute. Offering a completely independent architecture that duplicates every critical component of the primary architecture and provides multiple distribution paths, this fault-tolerant design provides twice the capacity required to operate at full IT load. This 2N, or N+N, model allows the facility to tolerate both planned and unplanned events, including scheduled system maintenance and unexpected outages and equipment failures. During a disruption, the redundant systems take over to ensure continued operations. With this level of redundancy, you will likely never be aware if an issue occurs at the data center.
For extra protection, a Tier IV data center may utilize a 2N+1 model. This provides twice the operational capacity (2N) as well as an additional backup component (+1) in case an additional failure occurs while the data center is utilizing the secondary systems.
The Tier IV data center list:
The Tier IV customer. This fully redundant design can be cost-prohibitive for many organizations, which is why this tier is generally populated by government organizations and large global enterprises with mission-critical servers and intense customer or business demands.
Lately there has been considerable talk about Tier V, which is not part of the Uptime Institute’s official Tier Classification System. Developed by a colocation provider in 2017, Tier V meets the standards of the Uptime Institute’s Tier IV while adding specific sustainable performance requirements including the ability to operate without water; detect and respond to outside air pollutants; monitor energy systems; and utilize securable server racks.
The Tier V customer: Given its focus on sustainable practices, Tier V data centers generally attract large organizations that demand full fault tolerance and embrace environmentally friendly operations and policies.
The growing need for availability is a driving force behind organizations’ reliance on third-party data centers. The uptime and performance delivered by these facilities allows you to build your business without fear of a system failure or natural disaster derailing your accomplishments. As you contemplate your reliability demands, the Tier Classification System can provide guidance on the data center tier level that can best balance your risk tolerance and budget.
SVP, Data Center Operations
Anthony Hatzenbuehler, SVP of Data Center Operations, is responsible for onsite data center operations activities, client services and the Operations Support Center (OSC).Read more from this author