Anyone involved in data center management knows there’s no such a thing as ‘business as usual’. Data center management is never simply a maintenance job – growing demand means that every move in the data center is critical. So, as the cost of network downtime increases and the data center plays an increasingly important role in the success of an organization, data centers must continually evolve to meet requirements.
To address these challenges, CTOs and CIOs are re-evaluating their IT and networking strategies, paying particular attention to their footprint, deployment options, and resource allocation. Priorities include deploying faster and more efficient optics, switching from the traditional three-layer vertical hierarchy to flatter, heavily meshed leaf-and-spine architectures, and migrating to higher lane speeds. However, the data center landscape is changing so rapidly, that data center managers often find themselves reacting to events and crises rather than implementing proactive strategies, which is absolutely necessary to make sure that the data center is prepared for future challenges and developments.
So, what should data center managers prioritize as they formulate a proactive strategy for the evolution of the data center? There are three key areas that, we believe, should be part of a successful strategy:
This post outlines considerations around migration to higher speeds.
It’s crucial for data centers to be able to support faster speeds throughout the network, as people, businesses and the world are increasingly hyper-connected and “always on”. Data centers must be able to support 25G/40G/100G and beyond – with three, four and five 9s of availability. Currently, data centers tend to use 10G links within the data center racks; 40G links are used between the racks, and 100G is typically used for long spans and connected to the wide area network (WAN). While the accepted migration path was 10G-40G to 100G-400G, the creation of the 25G Ethernet Consortium means that managers are rethinking their initial commitment in 40G and showing interest in 25G and 50G speeds.
However, regardless of speed, the most crucial consideration for data center managers is identifying the right type of infrastructure to support continued growth in link speed. This decision must take into account the following considerations.
As data centers are rapidly migrating from the traditional three-tiered design to a flattened two-tier fat tree design that increases redundancy and decreases latency, the network topology contains an increasingly high density of fiber ports in the spine switch layer. This model is changing the overall network architecture, encouraging the use of more space-saving fiberconnectors and modules, and is leading to the adoption of a structured cabling approach rather than a point-to-point cabling approach. This is because, while a point-to-point system often appears easier at the outset, it soon becomes unmanageable as the number of switches and assets grow. Ultimately, a structured cabling approach ensures scalability and can lower the data center’s CapEx and OpEx in the long term.
Singlemode fiber has been a tough sell for data center applications due to the high-priced optics required. Multimode fiber (MMF), on the other hand, has continued to offer a more attractive balance of performance, density and cost for enterprise data centers. Although distance can be a challenge for MMF, emerging high-quality components and engineered links can provide the link capacity to support longer distances and new data center topologies.
A new option has also emerged that may provide the optimum solution for fiber migration: wideband multimode fiber is a new fiber type, recently approved under ANSI/TIA-492AAAE, classified by ISO/IEC as OM5, and expected to be recommended by ANSI/TIA-942-B. CommScope introduced the technology in 2015 and it extends the capacity of laser-optimized 5 μm fiber such as OM4 by using wavelength division multiplexing (WMD) for optimized operation at 850 nm and beyond. By multiplexing four wavelengths, one strand of OM5 can, over a practical distance, increase data capacity by a factor of four.
When planning and implementing a migration to higher speeds, automated infrastructure (AIM) systems are becoming more widely adopted. AIM systems, such as CommScope’s imVision®, can assist in the migration process, using intelligent hardware and software components to map the identity, location and status of every managed port in the data center. This cabling infrastructure information enables data center managers to see and manage the connected environment in real time, discover networked devices, and pinpoint their physical location. This information is then used to help plan and execute the migration strategy.
Proactively addressing the key issues facing the data center is crucial in order to create a data center that is efficient, manageable and scalable. Data center managers must be able to manage increasingly complex infrastructures to pave the way for higher speeds and lower latencies. These strategies and considerations can help managers plan for a connected and efficient data center.
You can learn more about considerations for the modern data center here.