Before we talk about the levels of reliability of the data center (Tier Data center), would like to ask a simple question to readers. What is the idea of classifying data centers by level of reliability? And I personally have had a simple analogy – classification of hotels by stars. We, as consumers, guess that can be expected or, on the contrary, do not expect from a hotel with a certain level of stardom, when we choose a hotel to stay. Hotels are classified by 5 background levels, and data centers, at the moment are classified by 4-rep levels of reliability – Tier 1, Tier 2, Tier 3, Tier 4. As well as the stars of the hotel, the higher the level of reliability of the data center, the more reliable the data center is.
And who should determine the level of reliability of the data center?
The proposed tool for assessing the reliability of the data center in the form of certain parameters and requirements to engineering systems, to evaluate and determine the level of reliability of the data center is not only consumers who are planning to put the equipment or use the services of a data center, but also the investors who decide about investing their capital in the construction of the data center.
Development of tier data center architecture design
The idea to determine the levels of reliability of data center was born in the depths of associations of undertakings, gathered under the banner of Uptime Institute, which is engaged in the collection and processing of information, the issuance of the recommendation and the detailed requirements for the level of reliability of data centers.
TIA EIA 942 Standard
Each of the selected levels of reliability in the TIA EIA 942 standard, a detailed description of the requirements and recommendations of the following systems and components includes: architectural solutions, power supply, cooling, security, fire protection systems, structured cabling, conduit system, telecommunications. For example, the standard describes the requirements and recommendations for the minimum height of the raised floor for a certain level of reliability.
The standard in the description of the requirements for the amount of resources used by the letter «N» (short for need) and simple mathematical formula with the operations of addition and multiplication. The operation of addition is an increase of one standard unit (eg, N + 1 means that you need to have a stock of resources in one unit), and the multiplication operation is an increase by several times (for example, N * 2 is full duplication needs).
The first level of data center reliability – Tier I
Basic level of data center reliability. This level is used for data centers in the 60th and 70th years of the last century. Errors and failures in the systems and equipment at this level lead to a malfunction of the entire data center. Also, the data center is interrupted for maintenance and repair works. In the data center, there cannot be raised floors, redundant power supplies and uninterruptible power supply (UPS).
Engineering infrastructure created just to meet current needs, that is non-redundant and redundant resources (meeting the needs expressed in the form of the letter «N»)
Second level of data center reliability – Tier II
Data centers on the second level have a small level of redundancy health systems and have little excess capacity in engineering of the systems. But still susceptible to outages due to planned and unplanned failures of the equipment in the data center. For this level, you must have a raised floor, redundant power supply. Conducting technical and repair work will require a stop in data center.
The system has full redundancy, but additional elements are installed in the cooling systems of power supply and data centers (meeting the needs expressed as a formula “N +1 “)
Third level of data center reliability – Tier III
Data center with a given level of reliability allows for repair and maintenance work without stopping the data center, for e.g., it is possible to operate and replace the components of the system, adding and removing of faulty equipment. To ensure that the third level works well, it is necessary to design and build two pipelines to provide spare capacity operation of all equipment including failure or prevention of the electricity system for the cooling. But the mistakes and failures in the work may cause disruptions in the data center.
It has several paths (channels) for power and cooling distribution, but only one of them is active, has redundant components (software requirements expressed as a formula «N +1»)
Fourth level of data center reliability – Tier IV
Fault-tolerant data center with redundancy of all systems, allowing to perform all scheduled and unscheduled work without interrupting the data center. This provides the protection against failures. To meet the requirement of the fourth level of reliability to duplicate all systems given that in each system and its “Backup” will be having at least one additional component providing redundancy “N + 1″. That is, in the data center, there should be the level of system redundancy «N + 1» and the system itself still has to be at least duplicated. Failures may occur in cases of manual emergency disconnect of the power supply system and operation of fire safety. On the fourth level, even structured cabling system should be fully reserved.
Systems have dual redundancy, taking into account at least the additional part has a plurality of active paths and cooling load with 2 redundant components (N +1), i.e. UPS 2 N + 1 redundancy each (software needs expressed by a formula “2 (N + 1)”).
We should not forget that during the operation of the data center and the addition of servers and storage equipment in the data center with an unchanged basic needs of the engineering infrastructure in the data center that will grow and this can lead to a change in the level of reliability of the data center. That is necessary to review the level of reliability of the data center or in other words – to prove their “stardom”. But I doubt that a change of level of reliability inform the owner of a commercial data center – it’s not in his interest. Well, we all, as consumers realize that there is little to get a 5 * hotel or data center of third or fourth level of reliability (Tier 3, Tier 4), you must also properly exploit data center. While the standard does not address the reliability of the data center, depending on the level of exploitation.
Likely to adopt fifth level of reliability data center (Tier 5) with a coefficient of fault tolerance as a 5-nines 99.999%, which will meet the needs of modern businesses.