The enduring importance of data centre design

Alberto Zucchinali of Siemon highlights key data centre design points that managed service providers should follow so IT infrastructure meets clients’ shifting demands of the modern 24/7, 365 business.

  • Monday, 15th March 2021 Posted 3 years ago in by Phil Alsop

It can be challenging to support clients’ shifting requirements at the best of times, but recent months have raised the bar further. Changing work patterns and business processes have placed huge pressure on IT infrastructures and have driven additional demand for rapid deployment, scalability and reliability.

Whilst the benefits of using streamlined managed services to support organisations with critical IT capabilities - such as disaster recovery and improved network reliability - may have once seemed like a luxury for some businesses, more are now turning to managed services providers (MSPs). In fact, 67 per cent of companies now say they might explore MSPs.

MSPs however, especially those working within the data centre environment and who provide network infrastructure as part of their service portfolio, require the right IT foundations to meet clients’ rising demands in these challenging times. Part of these foundations is a well-thought-out data centre design and the correct selection of physical layer components. Only if a data centre facility is properly designed from the ground up can it effectively meet current and future business needs.

Design for scalability and rapid deployment

The overall rise in demand for computing power means that MSPs must be able to quickly expand their facilities and/or add additional capacity as required. A modular POD-based data centre design approach can facilitate a speedy data centre expansion process. PODs consist of groups of cabinets and are typically designed based on capacity, function or application. Once defined, they can serve as a template for incremental build-outs of additional units. As demand increases, this approach allows this design to be easily repeated as the data centre grows.

Faster infrastructure deployment can also be achieved by selecting the right cabling components. Rather than relying on field-terminated solutions where cable has to be pulled into place, cut to the required length and then terminated onto the connectors one by one, pre-terminated copper and fibre cabling simplifies this process. Pre-terminated trunking cables are factory-assembled and therefore are up to 75 per cent faster to install compared to traditional field-terminated solutions. They also offer guaranteed performance levels because they are manufactured and tested in a controlled and clean factory environment.

Design for speed

IT networks must also facilitate higher speed data transmission with speeds now swiftly moving from 100 to 200 and 400 Gb/s in the data centre backbone and from 10 to 25 and 40 Gb/s at the data centre edge. Selecting the right network cabling solutions on day one will determine if a data centre will be able support higher performance needs in the future. Future-proof network cabling systems such as Base-8 systems with 8-fibre MTP parallel optics connectivity for example provide the most efficient, cost-effective and highest performing option for current 8-fibre 40 and 100 Gigabit applications, and also provide an easy migration path to next generation 200 Gb/s and 400 Gb/s applications.

At the data centre edge, deployment of high speed interconnect solutions (HSIs) is highly recommended since they support server interconnect speeds beyond 10 Gb/s with low latency and low power consumption levels. These point-to-point cables facilitate speeds from 10Gb/s to 100Gb/s providing optimum support to customers who are looking to upgrade their network equipment to support 25 and 100Gb/s. High speed interconnects are typically available as Direct Attach Copper Cables (DAC), Active Optical Cables (AOC), or structured cabling using fibre assemblies and separate optical transceivers. Recent studies have shown that HSIs, particularly the AOC type - thanks to their closed assembly - can significantly contribute to reduced power consumption in the data centre and typically cost half as much as transceiver (TVR) assemblies.

Design for manageability and flexibility

IT infrastructures must remain flexible to support ongoing growth and data centre architecture plays a crucial role in keeping a data centre flexible and manageable as it develops over time. An any-to-all structured cabling design – as recommended by the EN-50600 data centre standard - allows for gradual data centre growth and simplified management. This is because in such a design all equipment is connected via ‘fixed cabling’ to one or more entirely passive zones, which are then used for easy and secure daily

patching, i.e. network moves, adds and changes. These centralised patching areas also allow servers to be placed where it makes the most sense for power and cooling, without the distance limitations of point-to-point cables and without the concern for switch port usage and availability.

Design for operational efficiency

Quality data centre design can also lower costs (e.g. for power) and improve operational efficiency (e.g. better PUE scores). Simple infrastructure design considerations like installing blanking panels in racks and cabinets for example can prevent the mixing of hot and cold air to improve cooling effectiveness and thermal efficiency, whilst the deployment of intelligent power distribution units (PDUs) which monitor power usage of individual connected devices determine if the devices operate within the intended range for energy use.

Again, the right choice of cabling can also aid airflow for improved cooling and thermal management. Patch cords with a reduced outer diameter (e.g. only 4mm) help maximise airflow for improved equipment reliability. Similar results can be achieved when high speed interconnect cords are selected according to the distance that they have to cover (e.g. when connecting a server down in the rack to the switch in a top-of-rack switching topology). Deploying cords that come in ½ metre increments will allow for gradual movement down to the bottom of the rack for equipment connections and will avoid high amounts of cable slack leading to improved cable management, airflow and a neater look.

Design considerations at the physical layer will enable MSPs to keep up with the growing demands and speedy changes of their clients’ business needs. In this context, detailed IT infrastructure planning and a careful selection of physical layer components including cabling, racks and cabinets, and PDUs is critical and will ensure that a data centre remains flexible, scalable and manageable in line with what the business demands from it.