Building a datacenter implies several challenges that must be well planned before starting. How can we plan a proper datacenter design if everything is changing every day?
Frequently, is used a methodology to design, build and plan the datacenter. Power/Square meter, the cost to buid/Square meter and tier level are examples of this methodology.
But these criteria can be misaligned with the goals, allocating resources to where they are not needed, leaving other areas with poor capacity. In addition, it will probably increase the costs of maintenance.
1. Poor planning for future needs
It is very difficult but one of the most important things that we need to define is the future capacity to expand.
It’s normal to plan a datacenter design that, in a few years, has no capability to expand or receive new equipment. The infrastructure should be designed thinking that will house several generations of IT equipment, AI, Big data storage.
The Power upgrades and cooling increments must be considered for future expansions and prevent the datacenter to become a dead end. Keep in mind that the average age of a datacenter is, according to IDC, 9 years and in this time everything can and will change.
In this stage of the technological evolution of IT, planning for future needs it’s almost impossible.
In other hand, the expansion capability that is not being used, increases the costs of ownership and, in some cases, will never be used.
2. Don’t let performance characteristics for last
The first thing to consider when drawing the datacenter, is the performance characteristics. Then the datacenter design criteria must take into consideration and last, the should be choses according to the design of the datacenter.
It is very important to start the project with a detailed floor plan, including office space, electrical equipment, IT staging areas, fire fighting systems and cooling equipment and also the necessary IT and communications equipment.
Then, with this floor plan, we can choose the site. Considering:
– The location (distance from end users normally means more latency)
– Power and communications availability
– Geographic hazards (floods, earthquakes, …)
– Proximity to services like fire stations, highways, …
Also, don’t forget to include the team that will perform the maintenance of all these systems and the Operations Team, in all these definitions.
3. Don’t Underestimate the Operations Team Training
The operation team will guarantee the performance level and keep the downtime to the minimum.
So, it’s very important to have this team in sufficient number and well trained.
Some datacenter owners tend to underestimate the Operations Team training and guidance. This leads to poor decision making in a stressful situation and less effectiveness when facing new challenges. Acting out some problem situations or unexpected situations periodically can be a good training and can help to prepare the team with the right tools.
The inclusion of this team in the datacenter design can avoid future renovations and design mistakes.
4. Ineffective Cooling process
Make sure that the temperature in the datacenter is planned correctly. The equipment doesn’t need temperatures under 20 degrees. In fact, if the temperature is comfortable for us, is also great for the machines. It’s more important to guarantee a good airflow than a very cold temperature.
The power consumption of the cooling system is around 40% of the total datacenter. And it will increase for under necessary temperatures and will increase the IT equipment power needs if the temperature rises over the standard.
A correct planning of the cooling system is very important to plan the space and the power needed for the site.
Related articles: Patch cord management: A Guide for better performance
