RSS
Логотип
Баннер в шапке 1
Баннер в шапке 2
2010/10/04 21:55:10

Certification of Uptime Institute Tier

The association of the enterprises, collected under the flag of Uptime Institute is engaged in determination of level of reliability of the data-center.

The directory of TAdviser of Data centers of Russia and technologies for data centers.

Content

TIER Standard of Topology Uptime Institute is an industry standard of reliability and fault tolerance of DPCs. It gives to customers all grounds for confidence that the DPCs which underwent certification of TIER will function definitely, quite in abnormal operating conditions, for example, after shutdown of power supply

For each of the selected reliability levels in the TIA EIA 942 standard the detailed description, requirements and recommendations is provided to the following systems and elements: to architectural concepts, power supply, cooling, security, fire protection system, structured cabling system, to the system of cable ducts, telecommunications. For example, the requirement description and recommendations about the minimum height of a raised floor for a certain level of reliability is provided in the standard.

In the standard at the description of the requirement to quantity of resources the letter "N" (reduction from the word need) and the simplest mathematical formulas with addition operations and multiplication is used. Addition operation designates increase in requirement by one conventional unit (for example, N+1 means that it is necessary to have a stock of resources in one unit), and multiplication transaction increase in requirements several times (for example, N* means complete duplication of requirements).

The 1st level of reliability of DPC — Tier-1

Basic level of reliability of DPC ((on classification of The Uptime Institute)). This level was applied to data centers in 60th and 70th of last century. Errors and failures of systems and the equipment at this level lead to failures in work of all DPC. Also work of a data processing center is interrupted for carrying out scheduled and repair work. In DPC can not be raised floors, reserve sources of power supply and the uninterruptible power supply units (UPSU). Engineering infrastructure is created only for satisfaction of current demands, i.e. without reservation and excess resources (ensuring requirements is expressed as the letter "N")

  • Downtime in a year — 28.8 hours
  • Coefficient of fault tolerance of 99.671%

The Tier levels began to be formulated by the American Uptime Institute in 1995, and a bit later – Telecommunication Industry Association in Europe. Determination of requirements to data processing centers in which crucial processes will be started was the purpose of these actions. At the same time developers proceeded from purely practical aspects. For example, for Tier I availability is performed just at the expense of the increased reliability of the engineering systems. Tier II means reservation of crucial nodes of infrastructure. In Tier III it is necessary to duplicate ways of receiving the electric power and coolant, and Tier IV demands full redundancy of all infrastructure and leaving from the city or federal power supply network as main source of energy.

Proceeding from statistical calculations and researches, for each Tier the availability indicator was defined. So, fulfillment of requirements of Tier I ensures functioning of servers in 99.671% of time, and Tier IV – at least 99.995%. However it is necessary to remember that all this is only statistics, and it has the same relation as existence of the coupon of checkup to the fact of functioning of the ABS brake system to real life.

The 2nd level of reliability of DPC - Tier-2

Data centers on the second level have the small level of reservation of operability of systems and have small excess resources in the engineering systems of the data-center. But are all the same subject to interruptions because of planned and not planned failures in the functioning of the equipment in data centers. For this level it is necessary to have a raised floor, reserve sources of power supply of DPC. Carrying out technical and repair work will demand a stop of work of a data processing center. A system has no full redundancy, however additional elements are installed in cooling systems and power supply of DPC (ensuring requirements is expressed as a formula "N+1")

  • Downtime in a year — 22.0 hours
  • Coefficient of fault tolerance of 99.749%

The 3rd level of reliability of DPC — Tier-3

Even the DPC constructed in due form can be insufficiently reliable because of unqualified operation, ignorance by engineers of regulations of service. Therefore Uptime Institute enters the concept Operational Sustainability ("operational stability") considering the general principles of the organization of work of a command on operation of data center: determination of load of personnel, relevance of regulations, determination of areas of responsibility in case of accidents and strict following to instructions.

The data center with the specified level of reliability allows to carry out maintenance and repair works without stopping of work of DPC. So operation and maintenance of a data processing center up to replacement of components of a system, adding and removal of the failed equipment is possible at the same time. To provide the 3rd level it is already necessary for the cooling system to design and construct two pipelines, to ensure reserve capacities functioning of all equipment taking into account failure or prevention of a power supply system. But errors in work and failures can cause interruptions in work of data center. Has several ways (channels) for distribution of power supply and cooling, but is only one of them active; has redundant components (ensuring requirements is expressed as a formula "N+1")

  • Downtime in a year — 1.6 hours
  • Coefficient of fault tolerance of 99.982%

The 4th level of reliability of the data-center — Tier - 4

The failsafe data center with reservation of all systems allowing to perform any planned and unplanned works without interruption of work of DPC. At this level reliable protection against failures is provided. To meet the requirements of the 4th level of reliability duplication of all systems is necessary taking into account that in each system and its "backup copy" there will be, at least, one more additional component providing reservation according to the scheme "N+1". So in data center reservation of a system to standard "N+1" should be up and a system still should be duplicated, at least. Failures can take place in cases manual abnormal shutdown of a power supply system and operation of fire safety system. On the 4th level even the structured cabling system should be completely reserved.

Systems have double reservation taking into account, at least, additional component Has several active ways of load distribution and cooling with reserve components 2 (N+1), i.e. 2 UPSes with redundancy of N+1 everyone (ensuring requirements is expressed as a formula "2 (N+1)")

  • Downtime in a year — 0.4 hours
  • Coefficient of fault tolerance of 99.995%

You See Also