Top-tier data centers: answering frequently asked questions about Tier IV

A week ago, we talked about plans to build a new Tier IV data center and immediately received several questions about this level in the Uptime Institute classification. From the discussions in the chats, we got a full-fledged FAQ. So today I will dispel the most tenacious rumors about Tier IV and tell a little what requirements of the Uptime Institute we are taking into account in the project of a new data center. 







What does “maximum possible level” mean, have you come up with something new?



Uptime Institute standards are over 25 years old. The  Tier classification system has existed for so long



Certification of data centers for Tier levels is carried out under several programs:



  • Certification of design documentation (Design Documents) - auditors check a package of design documents for the main engineering systems: air conditioning, power supply. They also study documents on related systems, for example, fuel supply. 

  • (Constructed Facility) — - . - , .



    Design. 

  • (Operational Sustainability) — . , .



    Design Facility. 



There is also a Management & Operations program to check the operation. But this is not a certification, but an audit of a data center, so we will not dwell in detail. 



The data center level is laid down at the concept and design stage. Therefore, we begin to prepare for Tier IV certification at the stage of building design, even before the design of engineering systems.



Why do we talk so much about Tier standards?

The Tier system contains a list of requirements for data centers of different levels. But there is no specific explanation of how to do this, only the requirements for the reliability of the infrastructure. Uptime Institute writes: 

"The Tiers standards welcome innovative engineering and recognize that all data centers are different."


This means that there are several options for how to comply with the requirements. 



We at DataLine deal with the practical side: we honestly look at the best European data centers, take the best practices, carefully try new things and apply this in the design of our data centers. We share our experience, including at our Universities



We have gained such experience of certification according to the Uptime Institute standards:



  • 2014 - passed Management & Operations audit. 
  • 2015 - the NORD-4 data center received the Design certificate.
  • 2016 - certified NORD-4 for Facility.
  • 2018 - NORD-4 received Operational Sustainability certification.
  • 2020 - NORD-4 confirmed the Operational Sustainability certificate.


What's next:



  • 2020 - together with Rostelecom-DPC, we began construction of a data center in Ostapovskiy proezd and its preparation for certification for Tier IV.
  • 2020 - in the second half of the year we plan to submit the NORD-5 project to the Uptime Institute.
  • 2021 - we plan to certify NORD-5 for Tier III under the Facility program.


So there is always something to discuss in the practice of applying standards, we will be happy to talk in our Salad Cart



What is the main difference between levels?



I have already talked a little about the redundancy schemes specific to different Tier. 



Let's look at the comparison table in the standard:







This is how the levels differ in the minimum number of active components that support the load (they are denoted by the very letter N):



  • Tier I - N is used - the minimum amount of equipment for the operation of the data center, that is, there is no reserve.
  • Tier II - engineering equipment is N + 1 redundant.
  • Tier III - according to the N + 1 scheme, engineering equipment and distribution routes are reserved: power cables, routes, pipelines. 
  • Tier IV - if there is a single failure of any equipment, there are still N active components. 


But it's not just about enkas, especially in the case of Tier IV. The main difference between Tier IV is that it is the only tier with fault tolerance. It's called Fault Tolerant Infrastructure. It also requires partitioning (or compartmentalization, I really like this word) and continuous cooling. Below we will see what this means.



Is Tier IV different from Tier III in the 2 (N + 1) equipment redundancy scheme?



As we can see, no specific Tier IV reservation scheme is specified. How to achieve N after any failure, each data center decides for itself. In the past, many took the Tier IV requirements too literally and proposed complex schemes like 2N + 1 or 2 (N + 1) to be sure to avoid failures. But in practice this is not necessary. 



What is Tier IV Resiliency? How is it different from Tier III?

In a Tier III data center, we tolerate failure situations where employees must step in and manually switch between redundant elements.



In Tier IV, such switches are absent or occur automatically. 



What is continuous refrigeration in Tier IV?



Tier III formally allows for a slight increase in the temperature in the turbine room when equipment fails and we switch between the main and backup air conditioner or chiller.



In Tier IV, this is much more closely monitored. Already at the design stage, it is imperative to provide calculations of the rate of temperature rise and prove that even theoretically it will not get hotter in the turbine room. 



What does it mean “in Tier IV systems are not only reserved, but also protected from physical impact”? How does it differ from Tier III?



At the fourth level, there should be a mandatory "partitioning" or "compartmentalization" of distribution paths. To fulfill this requirement, the data center needs to think over such a scheme so that the distribution paths take place in different rooms or in closed fire-resistant boxes. They will meet only in the turbine room. At the same time, the possibility of their maintenance and replacement should remain. 



If we fulfill this requirement, then we are not afraid of physical impact: even if the hammer falls on the cable, we will keep all distribution routes to the equipment in a separate place. 



What if there is a fire? 



All engineering equipment and all routes are located in separate rooms or protected boxes. This means that when one route is ignited, only the reserve is lost, without disrupting the operation of IT equipment.



And if a meteorite falls?



Our country is so harsh that sometimes it happens. Still, a data center is not a military facility to defend against such accidents. It's a shame, but the Tier system does not consider protection from a global flood, alien invasion, or reptilian takeover. 



Tier IV - is it twice as expensive?



Yes, somewhere it can get more expensive. But with the right approach, you can save money without losing quality. 



This is our first experience with Tier IV. A little scary, but we are moving in this direction. As soon as news comes out, we will be happy to tell you more. 



All Articles