Why high density is not always the best approach

Why high density is not always the best approach

High density racks can cause problems.

Technological innovation is all about doing more with less. Computers – which used to take up an entire room and required multiple staff members to operate – now have many more capabilities than their predecessors and have been scaled down to the point where they can be easily carried to and from the coffee house. 

Although this rule works well for much of technology's advancement, there is a limit to how much can be done in the data center with the area provided. High server density may allow for a more efficient use of space, but there are some major concerns that go along with it. 

" Racks using between 8 kW and 15 kW are considered high density."

What is high density in the data center?

The problem with discussing this topic is that different facilities are going to have their own special definitions on what counts as a high-density rack. With this in mind, the Association for Computer Operations Management (AFCOM) decided to step in and create a solid metric to define the term. Racks using between 8 kW and 15 kW are considered high density, while those using 16 kW or more are defined as being extremely dense. 

This is a great place to start in terms of narrowing down how many racks can be considered high density. This metric allows data center administrators the opportunity to track rack density through the use of power monitoring tools like those from Geist, thereby getting a more informed view of the current state of the facility. 

Overheating and overloading are major concerns

The reason that high and extreme density racks need to be discovered has to do with the amount of power that these collectives consume. Putting more servers in a tighter group may enable a better use of data center area, but it also consolidates the flow of electricity to a singular space. This can cause many problems, starting with the risk of overloading the circuit. 

Balance is a huge part of power distribution in the data center. Making sure machines are equally serviced not only works toward increasing energy efficiency, it also detracts from the possibility of a overworking the system. Specifically, too much power sent to a specific rack can end in a blown breaker. While the risk of downtime in such a situation is minimal if the facility has proper backup procedures, overloading the circuit is still a pain and might require some time-consuming intervention. 

Another problem that high-density racks can cause is overheating. Much like overloading, this problem stems from a large amount of electricity being utilized by a small portion of the data center's overall area. However, the power itself isn't the problem here. Rather, it's the heat generated by the energy running through the servers. 

Having a specific hotspot in the data center results in two outcomes: either the equipment overheats and fails or – more likely – the cooling system is forced to work harder. If the second scenario happens, officials can expect to see lowered energy efficiency due to increased cooling needs.

Combating an overheating event in a server rack can be costly to cooling efficiency. Overheating demands more cooling, thereby decreasing energy efficiency.

SMBs just can't handle the risk

Although these problems are apparent in data centers of any scale, small- and medium-sized businesses (SMBs) are particularly unprepared for mitigating the risks caused by high-density racks. The main reason behind this could very well be connected with how small IT departments are in these kinds of companies. 

The 2015 State of SMB IT Infrastructure Survey conducted by ActualTech Media discovered that SMBs often have extremely small IT teams. The study of 578 companies found that more than half had between 101 and 500 overall employees. However, a majority of respondents only had between one and four IT workers

An overloaded circuit or an overheated server can cause serious downtime, and might require physical intervention from an employee. This study shows that if one of these events were to occur, an SMB's IT staff could easily be overwhelmed. Even a best-case scenario involving the successful mitigation of such a situation would require a small one- or two-person team to drop what they're doing in favor of fixing the problem at hand, pushing back internal goals and previous projects in favor of ensuring uptime. 

High density in the data center may have some advantages, but the risks involved might be too much for many companies. Those organizations working with smaller IT departments should endeavor to steer clear of putting too many servers in a small space, as their team just may not be equipped to handle a failure.

Share