Subscribe

How efficient is your data centre?

Energy efficiency is relevant across the business, but particularly in the data centre where cooling is one of the biggest consumers of this precious resource.


Johannesburg, 14 Dec 2017
Devaksha Maharaj, System Engineer, Rittal.
Devaksha Maharaj, System Engineer, Rittal.

Energy efficiency, sustainability and green IT are just some of the newer challenges that IT administrators are faced with. Energy consumption has become one of the biggest cost factors in day-to-day operations, and so must be strictly controlled. Reducing electrical consumption will allow your data centre to be more sustainable and increase ROI. Continuous monitoring of data centre efficiency is essential if one wants to reduce one's power cost.

At the same time, cost-cutting must not come at the expense of the data centre's ability to serve its purpose effectively. A data centre is required to provision IT services to end users rapidly and to guarantee sufficient performance. Most modern data centres are characterised by high availability and resilience. In order to ensure this, all components and parameters of the building, server room and data centre that are relevant for security and availability must be monitored.

IT personnel are faced with the never-ending task of optimising data centre operations in terms of cost and environmental sustainability. One way to assess power usage by the data centre is the Power Usage Effectiveness (PUE) metric defined by the Green Grid. The Green Grid is a non-profit association of technology providers, end users, facility architects, utilities companies and policy makers. It works together toward improving the resource efficiency of data centres and IT globally.

There are five key steps that can followed in sequence to maximise data centre efficiency, although one can include an additional step - that of monitoring and measuring power usage so that improvements can be monitored along the way.

1. Optimising IT power

Depending on the level of availability and resilience required, power is supplied via one or more independent feeds. Since IT systems are what ultimately require power delivery, IT personnel need to try to lower the amount of power to IT equipment that's required. Sixty percent of this power is consumed by servers so taking the remaining four steps to reduce the power they require is crucial.

2. Optimising data centre space

The most important issue to consider is how a data centre's energy footprint (and therefore its cost structure) can be improved. The first item on the agenda is consolidation.

Consolidation can be achieved with the help of virtualisation technology. Older server models are replaced by the latest high-performance servers. Virtualisation software is installed on new servers, enabling multiple programmes with different operating system requirements to run on them.

While only a small fraction of the former systems' total capacity was utilised, the new, consolidated systems boast a much higher utilisation rate. Consolidation is clearly an effective way to drive down costs by reducing the number of servers employed (and the amount of power they consume). In addition, each server is used more efficiently than before.

If, as a result of consolidation, the servers' power consumption decreases, the data centre's total power consumption will also be lower, since less energy is used by the servers, and the cooling systems' output can be reduced accordingly. It is essential to determine the ideal operating point, so that no more cooling is provided than is actually required.

To improve the overall energy footprint, data centre infrastructure management (DCIM) keeps track of the IT infrastructure, the servers, and all relevant consumption data. Server management tools monitor the servers, their loads and the services running on them. Thanks to virtualisation, services can be allocated and scheduled with a view to minimising operating costs. For example:

* Some services can predominantly run at night.
* Weekend operations can focus on selected enclosures, while the rest of the IT infrastructure is run at lower load.
* Services can be moved to enclosures (or even data centres) whose cooling systems can exploit cost-effective free cooling longer.

These methods enable data centre operations to be optimised in terms of environmental sustainability and operating costs.

3. Optimising data centre cooling

Almost all the electric power fed into the data centre is ultimately converted to heat by the IT hardware. This heat must be dissipated from the data centre, which is accomplished by generating and distributing sufficient cooling. Cooling can be produced in various ways. The most effective method and the most suitable combination of technologies are selected in accordance with the customer's requirements.

Electrical cooling equipment is a major cost factor for data centres. Permanent savings can be achieved through the use of smart, adaptive climate control strategies.

If the cost of cooling can be reduced for a given server load, this can significantly improve the ratio between the amount of energy consumed by a data centre compared to the amount of energy it delivers to computing equipment (PUE) and overall data centre energy efficiency. This can be achieved by using natural sources of cooling e.g. direct free cooling. To exploit free cooling (i.e. the use of outside air for cooling) for a greater part of the year, the temperature of the servers' intake air must be kept as high as possible. Aisle containment and sealing the space between and alongside the servers are efficient ways of preventing hot and cold air from mixing

A further imperative is to size the cooling systems according to actual need. The system (water inlet and outlet) has a high inertia, and in order to manage it efficiently, it is necessary to measure the temperatures inside the enclosures (both in front of and behind the servers). By regulating the generation and distribution of cooling strictly according to need, smart IT management software can deliver additional savings.

4. Eliminating data centre power and cooling inefficiencies

Outdated power delivery systems, for example uninterruptible power supplies (UPS) and power distribution units (PDUs) can negatively impact PUE ratios. Evaluating the current situation, future requirements and modern alternatives require time and investments, but typically generate a good return in terms of PUE ratio improvement and savings.

5. Utilising DCIM tools

IT infrastructure management software such as data centre infrastructure management is a management platform for all components in the physical data centre infrastructure. Administration, monitoring and control of: access, climate control, power supply and safety. It does not matter whether a single rack or one or more data centres at different locations are involved. A DCIM solution improves the efficiency of the data centre.

Share

Sponsored Content

Click here to read more about the Green Grid data centre power efficiency metric. https://www.thegreengrid.org/