The technology industry is built upon standards that define and encourage interoperability. These standards create consistency in order to allow very difficult and diverse information to be conveyed to potential customers, IT staff, and other individuals who are interested in benchmarking solutions for comparison. There are different entities that are responsible for maintaining the requirements and guidelines for the standards.
The Uptime Institute is responsible for defining the tiered data center standard and they are the recognized authority for Tier certification. More importantly, by following the Uptime Institute’s stringent set of requirements and certification process executives can rest assured that a facility can meet availability requirements as evaluated by the vendor-independent third party. Data center tier or criticality is measured at four levels: 1. Tier I (Basic Site Infrastructure); 2. Tier II (Redundant Site Infrastructure Capacity Components); 3. Tier III (Concurrently Maintainable Site Infrastructure); and 4. Tier IV (Fault Tolerant Site Infrastructure).
So, let’s take a look at what it takes to meet the Uptime Institute’s standards. First of all, there is no checklist an engineering group can use as a formulaic cheat sheet. Instead, there are lists of required guidelines that data center designers must employ in order meet the Institute’s standards. The use of guidelines enables the Institute to remain vendor agnostic while conveying the necessary requirements. At Terenine, we are primarily interested in meeting the Tier III standard, and in this article we’ll focus on that. The Uptime Institute’s white paper, “Data Center Site Infrastructure Tier Standard: Topology,” specifies the requirements data center designers must follow and so we’ll summarize the information.
A Tier III data center must have a concurrently maintainable site infrastructure. In other words, Terenine must be able to perform maintenance on any piece of infrastructure equipment without harming or hampering computer equipment operation. That sounds easy until we look under the covers at what such an infrastructure needs in order to fulfill that requirement.
Servers, network equipment, IDS/IPS systems, SAN systems, and SAN switches put out heat. They put out a LOT of heat. Naturally, these warm natured creatures prefer cool environments and tend to get angry if they’re residing in a hot data center (to put it another way, the systems crash and the phone at the help desk will ring nonstop). To aid in understanding the cooling requirements of the systems to be provisioned in a data center each has BTU (how much heat they output) specifications.
The manufacturers provide these specifications so data center designers can create a sensible cooling environment and when aligned with Tier III standards, the air conditioning system requirements must meet peak demand during any maintenance operation. A sensible cooling environment, which meets peak demand means that the data center was designed to meet a specific capacity for cooling and past that capacity, the data center is considered overloaded (and can’t meet Tier III requirements). The second part of that requirement is that the system must be able to provide sensible cooling during any maintenance operation.
So, let’s say that one of the data center CRAC (computer room air conditioner) units has a Freon leak. In addition to posing a potential ozone threat, the unit will need to be taken offline and serviced in order to restore it to a normal operating condition. Without another separate CRAC system which can fully meet the data center’s sensible cooling requirements, some or most of the server systems in the data center will need to be taken offline in order to ensure they don’t overheat and malfunction. According to Tier III requirements, taking the server systems or other IT systems offline because of maintenance on a single infrastructure system (in this case, the CRAC unit) is unacceptable. Instead, the standard compels the certified organization to have a second unit, which can completely handle the data center’s sensible cooling requirements.
It is easy to see that this approach to data center design can quickly get expensive. Especially, once one realizes that this redundancy requirement applies not just to cooling systems, but also to every aspect of the infrastructure critical in supporting availability to the IT systems. UPS (uninterruptible power supply) systems must be physically redundant and individually able to sustain the data center’s peak capacity power requirements. PDU (power distribution unit) systems must deliver power over redundant mutually exclusive paths in order to ensure maintenance on one PDU system cannot affect the other. Furthermore, primary power supplying all of the operations must come from separate providers. The Uptime Institute considers primary power to come from engine-generator (or powertrain) systems which may be operated indefinitely aside from the local utility power grid. In order to have separate primary power systems, the data center would have to have a minimum of two systems where either single system could assume complete responsibility for infrastructure operations.
Tier requirements are at best met by whatever is the weakest system in place. For example, every piece of equipment in an infrastructure may fulfill Tier III requirements, but the cooling system is overloaded. As a result, it is no longer a Tier III compliant data center; instead, it would be a Tier II data center. This can be a painful realization to organizations that have spent hundreds of thousands or even millions of dollars on data centers in order to certify them compliant to a specific criticality.
The Uptime Institute does not provide or certify any intermediate levels to their specified Tier criticalities (in other words, no Tier 2.5, or Tier 3.14159…, etc). This often causes data center providers heartburn after they have sunk millions of dollars into building an amazing infrastructure, only to discover they overlooked something and now it will certified at a Tier lower than they were expecting. Human nature, being the way it is, quickly leads their marketing departments to start referring to their data centers with nonstandard, beer commercial-like descriptions: Tier II Lite, Tier III Ultra, and Tier IV with Less Carbs!
One of Terenine’s core tenets is “Truth in Advertising.” We have based our data centers on standardized concepts and specifications. And, we believe others should do the same. Euphemistically describing a data center, systems operation, or single piece of operating equipment will potentially mislead those who place their trust in our multimillion-dollar data centers and data system operations. We make the commitment that our clients will always be well-informed of our abilities and operating capacities and at no point mislead to place faith in pretty advertising gimmicks. Our engineers always design and implement our facilities according to the highest accepted standards, and of that, we are proud. This is our standard and we make no excuse when naming our capabilities.