No matter how ‘cool’ your cooling solution is, the fact is that poor airflow management in your server room or datacentre will lead to increased energy usage and ‘hot-spots’. In a typical datacentre, around 40% of the power consumed is taken up by the cooling system itself and so any improvements you can make, will contribute towards lowering operational costs.
Poor airflow management practices can build up over time but fortunately, there are some relatively easy-to-fit solutions that can fix these and allow for some quick wins.
What is Airflow Management?
Most server room and datacentre operators standardise on their rack cabinets. There is a reason for this. Apart from looking great, and assisting server, storage, and IT networking asset deployment, it is also easier to manage and control airflow.
Airflow management is a process and one that requires monitoring via some form of environment monitoring. The process is concerned with how ‘cold’ and ‘hot-air’ moves with a server rack, aisle row, containment and the server room or hall itself.
How the cold air within an IT facility is supplied is very much dependent upon room size and the size of the IT deployment. Computer rooms will have one or more cabinets on a solid floor and use a wall mounted air-conditioner. Larger server rooms and datacentres will deploy a more complex cooling solution such as computer room air conditioners (CRACs), computer room air handlers (CRAHs) and even some form of liquid cooling arrangement for certain high-power racks. A raised access floor and vented tiles will almost certainly be used, and this can lead to further airflow management issues.
In lower power racks and cabinets, airflow management solutions can sometimes be viewed as ‘plugging-the-gaps’. In this case airflow to ensure cold air from the air conditioner or cooling system, does not mix with the hot exhaust air from the IT assets.
As power demands increase within a rack, so does the cooling need and the potential for poor airflow, air mixing and leakage through poorly sealed side-panels, front, and rear doors. The greater the number of cabinets, the greater the cumulative effect of poor airflow on energy efficiency and cooling effectiveness.
The Ten Most Common Airflow Management Problems
There are several potential problems related to poor airflow in a server rack and datacentre environment. These issues can be compounded by the overall room layout design, choice of cooling system and deployment of server racks and can lead to unnecessarily high operating costs.
- Hot Spots: these are areas within a server rack or room that results from a build-up of excess heat, generated by the IT servers and networking devices. If we take the recommended ambient temperature range for a server facility to be 18-25°C, then a hot spot can be anywhere north of this and sometimes as high as 30-35°C or more. Heat rises and these sorts of temperatures can be found in the higher levels of a server rack. Heat also degrades electronics and high temperatures will almost certainly reduce server reliability and potentially lead to failures.
- Short Cycling: occurs when the cooling unit cannot complete its cooling cycle. For a wall mounted air conditioner, this can mean that the air return to the AC unit is impaired or at insufficient pressure or speed. Where the space under a raised access floor is used to deliver cold air to the front of the racks above, short cycling can be a common problem linked to the use of poorly sealed floor tiles or fitted grilles. A negative pressure can occur within the plenum in the worst case scenario.
- Bypass Airflow: occurs when the cooled air does not flow through the server racks and aisle arrangement sufficiently to cool them. The cooled air bypasses its intended target areas and ‘leaks’ into other areas, potentially mixing with hot exhaust air and leading to latent cooling.
- Latent Cooling: reduces cooling efficiency and occurs when hot air mixes with the cooled air during its return path to the air conditioner. Not only is temperature affected but also humidity leading to potential condensation build-up on cooled surfaces.
- Air Conditioning Failures: all cooling units require regular maintenance and service visits. They are electro-mechanical devices and as such, the harder they work, the more the need for service work and the greater the potential for a breakdown. Airflow management solutions can help to ensure that the cooling units deployed are used within the operating specifications required to ensure that they will deliver their expected operational hours and design life.
- Room Design and Layout: not ever computer or server room is a standard square or rectangular shape. Sometimes there are irregular angles and corners and especially in areas that have been made to fit rather than built specifically to house an IT facility. Ideally server racks should be arranged in a row with cool air presented to the front of the racks and hot air routed via the rear of the rack to the input arrangement of the cooling unit. Where build-out and service rooms are linked to the server space, these should also be air conditioned, and fitted with doors to provide secure access to the server area and a means of sealing them off to improve air flow. The airflow into and within these spaces should be managed. Floor plenum design is another area to consider in terms of its height and usage for cabling and piping. Floor grilles must be installed in the right place and in sufficient quantity to supply cooled air to the racks and at the right airflow rate and pressure.
- Room Usage: in smaller facilities, a computer room may also be used for general purpose storage. Not ideal for any IT environment as general materials stored in a computer or server room or build-out area can pose a fire hazard, as well as disrupting air flow. In a larger server room, a centralised UPS system may be used to provide power protection to all the racks. The larger the UPS system, the greater the amount of heat load that must be taken into account when sizing the cooling.
- Rack Layout Design: the equipment installed within a server rack will generate some level of heat. Some devices more than others. High powered servers will generate more heat than a networking switch or uninterruptible power supply. UPS systems with their lead acid (or even lithium-ion batteries) are typically installed at the bottom of the rack due to their weight and the need to keep the batteries (especially lead acid) within a lower ambient range of 20-25°C.
- Emergency Air Conditioning: during heat waves or to cover the failure or service an existing air conditioning unit, additional portable air conditioners can be used. Whilst portable AC units do provide additional cooling, their efficiency is poor, and they should only be considered as a temporary measure. If the heat load within a computer or server room is too high for the installed air conditioning system, then it should be upgrade and replaced.
- Environment Monitoring: monitoring ambient temperature and humidity in a server room provides important information as to the effectiveness of the cooling system. Environment monitoring devices can also be deployed at the rack level and is recommended for high-power racks where there is a high potential for hot-spots.
More information on cooling costs in datacentre environments:
Airflow Problem Solutions
During the operational lifetime of a server room or datacentre, air flow problems will occur. It is therefore important to establish best practice when it comes to airflow management and to regularly review the following areas:
- Server Room or Data Hall: there are two way to monitor the effectiveness of a cooling system and air flow management strategy. The first is with environmental monitoring which will collect continuous temperature and humidity data. The second is through energy usage monitoring. Most datacentres will monitor their PUE (power usage effectiveness) and drill-down into their energy usage data if the PUE figure deviates from its expected norm.
- Cabinet Rows and Aisles: where there are several server racks and cabinets, hot and cold-aisle containment can be used to improve and manage airflow. Containment solutions can be made to order as bespoke designs to fit the spaces and provide the working space for the server racks, engineers, and technicians. For smaller installations simply orientating two or more server racks in the same direction can improve air flow management. It should be remembered that cold air falls and warm air rises. Under this principle, wall mounted air conditioners should be placed above the rear of the racks so that the cold air flows over the top of them and drops down. The warm air is collected in a return path and rises to the inlets of the air conditioner.
- Raised Access Floors: provide an efficient way to cool multiple server racks but require thought and care in their design and installation. Underfloor plenum pressure and unobstructed airflow is as important as ensure floor tiles are sealed, floor grilles are in the cold aisle only and any cable or piping runs through the tiles themselves are sealed correctly. A poorly fitted floor tile will ‘leak’ cold air into a room.
- Server Racks and Cabinets: probably the most common problem area for many IT operators. Standardising on a rack design and manufacturer can help here as can the use of good air flow management practices and solutions. Blanking panels are just one type of airflow management product and are used to fill unused front-of-rack spaces. Gaps between installed devices should be minimised to potentially zero millimetres. Side panels and doors should be correctly fitted and sealed with foam tape and additional air barriers installed as required. In high power racks, additional environment monitoring should be deployed to cover front and back, top, middle, and bottom monitoring levels for overall air temperature monitoring. This can sometimes be done with an intelligent PDU supplied with additional sensor ports.
- Air Conditioner Air Flow Rates: as an additional measure, it may be necessary to use an air flow or pressure sensor connected to an environment monitoring device. This will provide continuous data and could generate an alarm when the pressure fails outside a pre-set threshold level with emails and SMS txt messages sent out to a distribution list.
Air flow management has always been an important consideration for datacentre operators but has risen-up the IT management agenda for computer and server room managers. In recent years there have been more weather extremes to consider including high temperature summer heat wave and recently the rise in remote working which has left many sites reliant on limited on-site staff. Servers themselves continue to demand higher amounts of energy and IT asset upgrades can bring to light air flow management issues that were previously not an issue. There are several quick fixes including blanking panels that can be easily installed. Not only will these bring immediate improvements, but their use will also improve overall cooling system and energy efficiency.