The TIA-942 standard

Advancing in the topic of information security, we will talk in this new installment of the TIA-942 standard, which establishes the characteristics that must be executed in the components of the infrastructure for the different degrees of availability.

by Gustavo García Enrich

Within the it world are some intrinsic properties of information such as availability, which must be preserved to ensure continuity of operations and business. Also in this world converge some risk factors external to information, such as fire, which can destroy it and cause the organization great losses.

In previous articles we mentioned these two aspects and analyzed fire as a critical risk factor because it conspires against availability by attacking the supporting infrastructure, but in turn the infrastructure itself must work 7x24. Transferred this property to the field of action of the datacenter, it must be considered as the interrelation of a series of infrastructure subsystems that support critical equipment (hardware), to maintain an availability of systems adequate for the characteristics of the business in which we find ourselves.

Keep in mind that not all activities require the same level of availability and this will arise from a previous analysis called BIA (Business Impact Analysis) that economically quantifies the impact of a datacenter shutdown on the organization's business.

In general, we can establish a priori an approximate classification of the criticality of the systems for different areas of activity.

Infrastructure and the TIA-942 standard
In April 2005, the Telecommunication Industry Association published its TIA-942 standard with the intention of unifying criteria in the design of technology and communications areas. This standard, which in its origins is based on a series of specifications for communications and structured cabling, advances on the infrastructure subsystems generating the guidelines that must be followed to classify these subsystems according to the different degrees of availability that is intended to be achieved. In its annex G (informative) and based on recommendations of the Uptime Institute, it establishes four levels (tiers) based on the redundancy necessary to reach availability levels of up to 99.995%.

In turn, it divides the supporting infrastructure of a datacenter into four subsystems, namely:

• Telecommunications 
• Architecture 
• Electrical system 
• Mechanical System

Understanding the tiers
One of the biggest points of confusion in the field of uptime (available time of systems) is the definition of reliable datacenter; since what is acceptable to one person or company is not acceptable to another. Competitive companies with completely different datacenter infrastructures claim to have high availability; this may be true and will depend on the subjective interpretation of availability that is made for the type of business in which a company is located.

The truth is that to increase redundancy and reliability levels, single points of failure must be eliminated both in the datacenter and in the infrastructure that supports it.

The four levels of tiers proposed by the standard correspond to four levels of availability, having to the greater the number of tiers, the greater the availability, which also implies higher construction costs.

This classification is applicable independently to each subsystem of the infrastructure (telecommunications, architecture, electrical and mechanical). Keep in mind that the global classification of the datacenter will be equal to that of the subsystem that has the lowest number of tier. This means that if a datacenter has all tier IV subsystems except the electrical one which is tier III, the overall classification will be tier III.

It is important to take this into account because when the adaptation of current datacenters to tier IV is intended, in places like Latin America, there are physical limitations difficult to overcome in the current building sites. Practically to achieve a tier IV datacenter you have to design them from scratch with the standard in mind as a guide. A clear example of this is that it is very difficult to achieve the power supply of two independent substations or to be able to achieve the heights required by the standard in existing buildings (3 m minimum on raised floor and not less than 60 cm between the roof and the highest equipment).

The standard describes, in summary, the different tiers as follows:

Tier I: Basic Datacenter
A tier I datacenter can be susceptible to both planned and unplanned outages. It has air conditioning and power distribution systems; but it may or may not have a technical floor, UPS or electric generator; if you have them, they may not have redundancy and there may be several single points of failure. The maximum load of the systems in critical situations is 100%.

The datacenter infrastructure must be out of service at least once a year for maintenance and/or repair reasons. Emergency situations can lead to more frequent shutdowns and operational errors or failures in the components of your infrastructure will cause the datacenter to stop.

The maximum datacenter availability rate is 99.671% of the time.

Tier II: Redundant components
Datacenters with redundant components are slightly less susceptible to outages, both planned and unplanned. These datacenters have a false floor, UPS and electric generators, but they are connected to a single electrical distribution line. Its design is "what is necessary plus one" (N+1), which means that there is at least one duplicate of each component of the infrastructure. The maximum load of the systems in critical situations is 100%.  Maintenance on the electrical distribution line or on other components of the infrastructure can cause a processing interruption.

The maximum datacenter availability rate is 99.749% of the time.

Tier III: Concurrent Maintenance
The capabilities of such a datacenter allow you to perform any planned activity on any component of the infrastructure without interruptions in operation. Planned activities include preventive and scheduled maintenance, component repairs or replacement, adding or removing items, and testing components or systems, among others. For infrastructures that use water cooling systems it means double set of pipes.

There must be sufficient capacity and double distribution line of the components, so that it is possible to carry out maintenance or tests on one line, while the other serves the entire load. In this tier, unplanned activities such as operational errors or spontaneous infrastructure failures can still cause a datacenter outage. The maximum load on systems in critical situations is 90%.

Many tier III datacenters are designed to be upgraded to tier IV, when business requirements justify the cost.

The maximum datacenter availability rate is 99.982% of the time.

Tier IV: Fault tolerant
This datacenter provides the ability to perform any planned activity without interruptions in critical loads, but also the fault-tolerant functionality allows the infrastructure to continue operating even in the face of an unplanned critical event. This requires two simultaneously active distribution lines, typically in a system + system configuration; electrically this means two independent UPS systems, each system with an N+1 level of redundancy. The maximum load of the systems in critical situations is 90% and a level of exposure to failures persists, due to the start of a fire alarm or because a person initiates an Emergency Power Off (EPO) procedure, which must exist to comply with fire or electrical safety codes.

The maximum datacenter availability rate is 99.995% of the time.

To put into perspective the availability rate that is intended for the different tiers, Table 2 expresses its meaning expressed in the annual downtime of the datacenter. These percentages should be considered as the five-year average.

Keep in mind that for a tier IV it is contemplated that the only stop that occurs is due to the activation of an EPO and this only happens once every five years.

However, for the requirement that demands a tier IV some companies or organizations state that they need an availability of "five nines", this means a 99.999% availability. This is just over five minutes a year without systems.

Conclusion
The purpose of the TIA 942 standard is to provide a series of recommendations and guidelines for the design and installation of a datacenter. The intention is that it will be used by designers who need a finished knowledge of facility planning, cabling system and network design.

The TIA 942 standard and tier categorization are booming in Latin America. This is good because it leads to the rethinking of infrastructure needs in a rational way and aligned with the availability needs of the business in which organizations are located.