Guest post by Interxion: How are data centres classified against the Tier Classification System?
You may well have already come across the term "Tier" in connection with data centres. But just what really is meant by this? In data center geek-speak, it stands for "category of performance".
The tier system has been introduced by the American Uptime Institute to classify the availability of data centers. More recently, the Institute revised its certification process in July 2015, noting that there were data centers without official classification that claimed the Institute had certified them. Much of the controversy took place between Tier III and Tier IV rankings.
Tier I - the lowest category
Tier I describes a system with a 99.671% availability rate - data centers of this category are sufficient for smaller companies where IT is primarily used to manage internal processes and whose Internet presence is purely passive. They are also described as 'powered warehouses'.
Tier II – a lot is still being tolerated
The Tier II rating still allows many mistakes, including human errors, to be made in order to meet the 99.749% availability. Tier II is suitable for smaller companies whose IT is usually only used during business hours and who use their Internet connection for business purposes but only risk delays and data loss in the event of a failure. Any delays that occur do not harm the business itself.
Tier III – it's getting tougher now
With Tier III, the requirements become much stricter: the availability for Tier III is 99.982%. Redundancies, for instance, must be spatially separated and alternating maintenance must be possible, in other words: it must be possible to carry out maintenance work without affecting the continuity of the servers located in the data center. Tier III, for example, is intended for companies that are operational around the clock for their customers, but can also be reached and operated over other channels in case computer operation is temporarily unavailable.
Tier IV - the highest availability category
The difference between Tier III and Tier IV is actually minimal. In order to guarantee an availability of 99.995%, the data center must be completely redundant in terms of power supply, cooling and network compared to Tier III. This architecture permits even the most extensive technical faults to be rectified without interrupting the availability of the servers located in the data centre.
Tier IV is suitable for companies that need to reliably deliver a service 24 hours a day, 365 days a year to remain competitive: enterprises that rely on e-commerce, electronic market transactions or financial services; and global corporations that operate in multiple time zones.
For a Tier IV classification, a data center must meet the following requirements:
- Zero single points of failure: Tier IV vendors have redundancies for each process and data protection stream. No single failure or error can shut down the system.
- 99.995 % uptime per year to maintain ranking IV.
- 2N + 1 infrastructure (twice the amount required for operation plus backup). 2N + 1 therefore means "fully redundant".
- The maximum downtime per year must not be exceeded. Providers must plan for some downtime due to optimization work, but this must not affect operation.
- 96 hours power failure protection. This power supply must not be connected to an external source and must be completely proprietary.
High availability at Interxion
At Interxion, availability and reliability are paramount. Colocation in Interxion's state-of-the-art data centres offers a wide range of power supply options complete with backup, cost-effective air conditioning, advanced fire and water detection systems, a multi-level security concept, ISO27001 and BS25999 certified information security management systems and business continuity systems. Interxion delivers 99.999% Availability - even exceeding Tier 4.
Tier I to IV Classification: An "Exclusivity" of the Uptime Institute
The Uptime Institute created their tier classification system in the mid-1990s. "Tier", for example, was also used in the ANSI/TIA-942/TIA-942A standard of the Telecommunications Industry Association (TIA) in a slightly different context, giving rise to some confusion. Both institutions have concurred to correct this difference in interpretation - TIA has since dropped the word from its standard vocabulary and is now using the term 'Rated Levels'.
Originally published on 04.10.2018 in the Interxion-Blog.
Translated by Edina Gallos / aspectra.