The advent of cloud computing has placed the data centre at the core of many businesses today. Mission-critical applications demand highly resilient solutions, including cabling infrastructure, explains Luca Rozzoni.

The main purpose of a data centre is running applications that handle the core business and operational data of an organisation. The growth of the Internet, coupled with applications such as video streaming and ecommerce, have led to a huge increase in the amount of data that is being processed, fuelled by a global economy that requires data access at all times.

To support this, each component of a data centre and its supporting systems must be designed to work flawlessly together, providing the highest reliability levels. Most importantly, physical infrastructure has to be able to provide a seamless migration path able to support rapid changes in business requirements and embrace future applications.

The two most significant standards for data centre, ISO/IEC 24764 and TIA/EIA 942, both have a similar hierarchical structure for cabling subsystems.

Data-centre functional areas

Storage, main distribution and zone distribution are the supporting pillars of any data centre architecture. Different network interfaces and applications can be used to create connection between the various elements. For instance, Ethernet is adopted in the networking area contained within the main distribution area, whereas Fibre Channel (FC) and Fibre Channel over Ethernet (FCoE) support communications in the storage area. It is relevant to underline that these applications are not constrained by any physical media, so both fibre and/or copper can be deployed across the different areas.

Storage architectures

Storage has become probably the most critical and yet the most  vulnerable functional elements of an enterprise’s data centre. Traditional direct-attach storage (DAS) deployments have been preferred in the past for their low cost of ownership. However, since applications have become more complex and the need for flexibility has become more important, there has been a move towards more centralised approaches.

For this reason, network attached storage (NAS) or storage area networks (SAN) have become more popular – with the added benefit of reducing hardware and cabling infrastructure. These innovations are driving faster data transfer speeds such as 10/40/100 Gigabit Ethernet and 2/4/8 Gigabit Fibre Channel (and future 16 Gigabit Fibre Channel).

The right storage solution for an enterprise depends on the specific needs and long-term business goals of the organisation. Digital assets will continue to grow in the future, and it is therefore of paramount importance to make sure that the storage infrastructure can be scaled up to meet growing requirements easily and cost-effectively.

It is also important to consider structured cabling solutions that allow for easy migration and expansion of storage investment without adopting a “rip and replace” approach. The main goal of a sound structured cabling solution is to be able to support current applications while providing a seamless pathway transition to future technologies.

Network consolidation

Many data centres today are still operating multiple parallel I/O networks to support various applications: this is costly, and so there have been several attempts at merging these networks into one consolidated physical infrastructure using Ethernet as the main application. This makes even more sense considering the emergence of 40 and 100 Gbit/s data rates, which will allow Ethernet to be an effective platform well into the future.

The move to consolidate I/O networks has been the main driver for the development of Fibre Channel over Ethernet (FCoE). FCoE’s main use is in data-centre storage area networks (SANs) where it can reduce complexity of design and implementation. With FCoE, network (IP) and storage (SAN) data traffic can be consolidated using a single network, reducing the number of network interface cards and switches and reducing power and cooling costs.

Consolidation of I/O systems can bring significant saving in terms of equipment, rack space and connections. However, more data traffic is now running through the same cabling infrastructure, with the potential risk of creating bottlenecks and data loss. It is critical to take extra care in the selection of high-performance cabling solutions, which have to be capable of withstanding high volumes of data transmission in any given environment, while delivering the highest signal performance.

Main and zone distribution

In the ISO/IEC 24764 standard, the main distribution (MD) system houses the main cross-connect, core routers and switches. The zone distribution (ZD) system can be seen as the main transition point between backbone and horizontal cabling, and houses the LAN and SAN switches that connect to servers and storage devices.

As data centres continue to face the need to expand at a rapid pace, the fundamental concerns related to main distribution and zone distribution remain constant. A properly planned data-centre infrastructure must be able to deliver three key strategic concepts:

• Agility by providing optimum flexibility in design and implementation
• Availability of network under the most stringent conditions
• Efficiency through the highest and most reliable network performance.

It is essential that these concepts be considered during the planning, design and implementation of the project. This will help to preserve initial investment in the project, and it will guarantee the data centre’s capability to respond to future changes.

The projected growth of 10GbE, 40GbE and 100GbE underlines the importance of building a network infrastructure today that can support the changes in technology in the near future. 10GbE is still going to play a major role throughout the next five years alongside the rapid growth of 40GbE by the end of the decade. Keeping on top of these changes is a vital part of a company’s ability to retain its business success. Deploying 10GbE today becomes a strategic decision: not only is it necessary to support today’s application; more importantly, it also has the task of providing a reliable migration pathway to 40GbE and 100GbE.

The main goal of a successful migration strategy is to ensure the original investment in the network can be scaled up, and can accommodate the latest technologies. This task needs to be performed with minimum disruption and by delivering the performance requirements expected by the standards. One simple way to do this is by using pre-terminated cabling components, so that upgrades can be achieved by swapping only a few components while preserving the rest of the physical infrastructure.

It is also vital that today’s physical infrastructure is built using ‘best in class’ low-loss components and fibre cable. These must be designed to deliver the highest bit-error-rate performance with the lowest insertion-loss and return-loss values.

Conclusion

When it comes to the physical infrastructure elements, it is of vital importance to build a network foundation that can support short-term needs while keeping track of long-term objectives. Successful migration strategies are modelled around the concept of agility, availability and efficiency in order to maximise current investment, minimise risk and provide a robust migration path for the adoption of future technologies.