The need for speed in the datacentre

 

Huge growth in data volumes especially video, an explosion in the number of connected devices, plus the need to address IT imperatives, are escalating demand for more storage, processing power and network performance.

 

Datacentre managers are constantly searching for new solutions, allowing them to scale growth and performance - while reducing energy costs. For many it’s a matter of constant catch-up with their datacentre infrastructure, while others face the reality of quickly running out of space, capacity or power. When this happens the options available include upgrading facilities, building a new private datacentre, or leasing colocation facilities.

 

Regardless, fast and flexible provisioning has now become the prerogative. This means that the traditional, construction-based datacentre model with its long lead-times simply won’t be able to keep up. Wherever the applications are hosted and delivered the datacentre of today needs to be more efficient, scalable and able to respond quickly to changing business needs as well as cut down on capital and operating costs.

 

When it comes to the infrastructure inside the datacentre, it is the physical cabling systems that tie together all the switches, servers, storage and other IT assets. The cabling infrastructure needs to stay well ahead of applications, networking, server and storage equipment upgrades, else it becomes a big obstacle to deployment. Given that businesses need to be very competitive and agile, the key questions that need to be asked are:

 

1) What can the cabling infrastructure do better?

2) What does better look like?

3) How can it improve datacentre efficiency?

 

Firstly, datacentre project implementations increasingly have to meet tight deadlines and a rapid time to operation. In these situations it is important for manufacturers to provide short lead-times, so that infrastructure solutions can be expedited to improve project delivery and accelerate the availability of any new services and applications. With 24-hours-a-day, seven-days-a-week operation at stake, the need for a high-performing, reliable network is a must. Therefore datacentre operators, owners and managers should look for 100 percent factory-tested cabling solutions, which provide consistent quality to ensure maximum system performance.

 

Once the cabling hardware is delivered, a structured cabling solution needs to support the flexible provisioning of cabling trunks throughout the datacentre, enabling the IT team to respond quickly to changing business needs. New cabling trunks should be able to be quickly added, interconnected and easily routed to the new IT equipment racks without any disruption to existing operations. Installation and configuration processes should also be made simple to avoid the need for different components, configurations or lengthier moves, adds and changes. For instance, installation processes for cabling projects can be made simpler and faster through the provision of high-quality, factory-terminated and tested, modular cabling connections. Where fibre-rich cabling infrastructure is used, Universal Polarity wiring in the fibre optic modules and reversible adaptors can provide fully managed optical polarity of the fibre links that helps reduces moves, adds and change costs by up to 25 percent versus tool-oriented solutions.

 

A high-performance, reliable cabling infrastructure must support a high-availability profile, not only during installation, but over time as it grows to support new capacity and applications. The initial installation and any subsequent expansion must utilise high quality components and terminations to ensure maximum performance and uptime. Over time, after many moves, adds and changes, datacentre entropy builds up. As a result of entropy, fibre patching frames, cable baskets and ducts become overcrowded leading to tight bends, pulled connectors and trapped cables, which degrade optical link performance, or result in the loss of circuits. Therefore the use of low-profile bend-insensitive optical fibre cable is a must to beat entropy. Such cables exhibit up to a tenfold reduction in bend-induced loss over conventional multimode fibre, protecting power margins to ensure systems continue to operate and unscheduled downtime avoided.

 

Virtualisation technologies applied to higher performance server and storage clusters provide increased capacity and computing efficiency, but also drive a migration to higher networking speeds within the datacentre. Independent research suggests a 100 percent year-on-year growth for switches that enable 40G and 100G speeds, which will become the norm by 2017. The majority of spending in 2017 is predicted to come from systems offering 40G Ethernet for secondary network connections and 100G Ethernet for links within the datacentre and the network backbone. Looking further down the line, the standard for 400G is being worked on. Therefore any cabling infrastructure put in place today needs to be designed to support future applications and equipment deployment without disrupting datacentre operation, let alone ‘rip and replace’.

 

Higher speeds, such as 100G, use parallel optics that require the connection and use of four transmit and four receive fibres. A modular solution to supporting parallel optics, leaving existing cable hardware and trunk cables in place, becomes crucial.

 

The growth in 40G and 100G connectivity within the datacentre means that the fibre density will also increases exponentially. Traditional fibre panels accommodate up to 96 fibres in 1U of rack space, so any increase in fibre would require far greater floor space. To mitigate any fibre-fuelled explosion in rack space, structured cabling for 40G and 100G speeds needs to provide a much higher density of fibres per 1U.

 

Finally, the deployment of a fibre-rich cabling infrastructure also plays a significant role in contributing to the improved energy efficiency in the datacentre. The lower power consumption of optical transceivers and the fewer number of switches required in an optical system results in potential energy savings of up to 25 percent when compared to the copper equivalent. The high-fibre counts and low profile of optical cable also provides as much as 30 percent reduction in physical cable space, keeping void space clear of congestion with no cooling energy losses in the raised floors.

 

In summary, today’s datacentres need to be more efficient, scalable, and able to respond quickly to changing business needs. Datacentres all rely on the physical cabling infrastructure to support connectivity and application access. Such infrastructure solutions need to be designed to support a need for speed. The ability to get the cabling infrastructure installed and operational in the shortest possible time is crucial for the majority of projects. Rapid installation processes inherent to the solution and expedited delivery commitments from manufacturers ensure that these projects come in on time. Once in place, the cabling infrastructure must accommodate rapid provisioning and efficient scalability for future data demands and the changing needs of businesses. While the move towards higher performance and high speed connectivity requires simple, efficient, non-disruptive and cost-effective migration of the installed cable infrastructure.