Rittal’s Clive Partridge explains the key considerations for planning rack developments to ensure an optimal installation
There are a number of key areas for IT and data centre managers to consider when it comes to planning future rack developments. One of the most critical is deciding the precise configuration of the internal components. This typically underpins the success of the whole project, especially when IT racks are not installed in a standardised data centre environment.
Meanwhile, choosing which rack to install requires an understanding of its purpose, the routing of cabling for power supply and networks, and the ideal cooling solution.
Here is a summary of some of the key questions to ask to help ensure the success of any future installation and commissioning of racks.
What will the rack be used for?
The number and type of components will have an impact on the size of the rack. If it will simply be used to house servers, then an IT rack 600mm wide will suffice.
However, if it is primarily for network components, the rack should be 800mm wide to accommodate the cabling. But, increasingly, enterprises are combining server and network components within individual IT racks.
As IT takes on an increasingly significant role within organisations, more and more components are packed in to make best possible use of existing infrastructure. This means the largest available racks should be selected in line with space constraints. An IT rack that is 42 units high, 800mm wide and 1,200mm deep provides ample room for custom configurations and allows for future expansion.
What form of climate control is needed?
Will the rack be installed in a room that does not have an integrated cooling system? If so, then a suitable cooling system needs to be considered from the outset. If just a single IT rack is needed, then cooling could simply be roof-mounted fans. If a greater cooling capacity is required, then a compressor cooling unit can be added, either externally or internally, in the form of a DX system – although a higher IP rated rack will be required. DX systems are becoming more popular as they are relatively easy to install and can be deployed in office environments thanks to the low noise levels of the internal component.
Should the rack be deployed in a conventional data centre, with either perimeter cooling units or InRow cooling units adjacent to the racks, then perforated doors should be used. Additional benefits can also be obtained here by deploying aisle containment or cocooning, where either the cold or the hot aisle (or in some cases both) is enclosed, the major benefit being lower operating costs and more efficient cooling (ie better delta T – the difference in temperature between two measuring points).
Sealed Up?
There are various options for cooling IT rack interiors depending on their use. For example, inside server enclosures the cool air should flow from front to back, while in network enclosures it should be directed through the parts that need to be cooled.
In both scenarios, it is important to seal around the front 19” section to ensure cool air does not leak into the ‘hot zone’ at the rack rear and therefore lower the delta T across the rack. All open rack units should be closed off with simple 1U snap-off section blanking plates for effective separation. To this end, there is a range of accessories available to enhance cool-air routing, enabling horizontal airflow for side ‘breathing’ switches, and more. Effective seals and climate control solutions tailored to the IT rack’s specific purpose will translate into improved energy efficiency. A little bit of planning goes a long way when it comes to reducing energy costs for IT operations.
Where do you want your cables?
A detailed plan for interior and exterior cable routing should be in place before a server enclosure is purchased and configured. The power supply, in particular, has to be considered. Many active IT components require a redundant power supply which generally means two rack-mounted vertical power distribution units (PDUs) at the rack rear, left and right for an ‘A’ and ‘B’ supply respectively, making cable management a bit more complex. Moreover, best practice usually recommends power and copper data cables should be kept physically apart to avoid the unlikely possibility of electrical field interference. Special attention must be paid to the minimum bend radius of the fibre-optic cables to prevent signal attenuation.
If the racks are to be placed on a raised floor, then power and network cabling can simply be laid on separate cable trays underneath. You can also mount cables under the ceiling and route them above the top of the racks. In this scenario, the roof plate needs to be configured correctly, openings should be closed off with brush strips creating a tight seal, simplifying cable management and improving energy efficiency. It also means that even after cables have been installed, accessories (including roof plates with multiple parts) can be easily removed for simpler maintenance and retrofits.
Many rack vendors offer a choice of elements for effective cable management, enabling customised component configurations. Both open and closed cable duct systems are available for horizontal and vertical cabling, and cabling between thermal zones.
Typically, air-tightness and a defined air pressure must be maintained to prevent warm air and cool air from mixing. This important consideration is often built into the design of many rack vendors’ enclosure products. The enclosure supports the need for air circulation in conjunction with the corresponding cable installation components.
External cable management?
Crowded IT racks have little available space to squeeze in new components. One answer is to route cables outside the racks such that cable ducts pass through the side of the rack, routing cabling over the top so that it re-enters the rack from the opposite side. Although this approach can save a great deal of space, it may make identifying individual cables more difficult and maintenance work more complicated.
Wherever easy maintenance is a priority, cables should be routed inside the racks. Pinpointing and replacing a single cable is then far simpler if there is a failure or if a reconfiguration is required.
Recent developments
Innovative products are now becoming available for in-rack cable management. One of these seems quite interesting: the network cable organiser (NCO).
The NCO is a 482.6mm (19”) cable storage system. It takes up 1U in the network enclosure and contains 24 tested CAT 6, Class E patch cables or fibre-optic cables, each with a length of 1.6m (sufficient for 23U). It has a modular structure and is made up of individual cassettes. The pulley system integrated into the cassettes allows surplus cable lengths to be drawn in automatically which ensures that every cable is available in the perfect length, eliminating the need to order and stock a variety of different cable lengths. The result is permanently well-organised cable management, providing a perfect overview of what is connected where.
It is also more energy efficient, avoiding airflow blockages that can arise as a result of surplus lengths of the individual cables inside the enclosure facilitates more efficient cooling. And the NCO saves on space, because it only requires 1U per switch. Cable management panels are not required, saving additional space, while the flat cables used in the NCO only use one-third of the space of a round cable.
The flat cables still offer the same level of reliability, are 100% tested and the fibre-optic cables additionally have a measurement record on the cassette.