I think that’s a fair observation. Unfortunately, it generally only applies to some aspects. Cable networks often follow the same old patterns even if they are obsolete or inefficient for the customers’ needs. Some publications are misinterpreted, turning supposed innovations into disadvantages when installed. A typical misunderstanding would be that single-mode cables are better suited to higher bandwidths and therefore more future-proof. Subsequently, customers have single-mode fibre installed instead of multi-mode. In actual fact, however, manufacturers are making great strides in transceivers for both. The only generalisation we can make is that single-mode cables can be favourable at distances of 500 metres or more and are absolutely essential for longer distances of one or two kilometres and up. So the better choice of fibre really depends on the purpose at hand.
Have a look at their inspection reports. In addition to attenuation and length, other values should be included that confirm suitability for, say, data rates of 40 Gb/s. A specialist should be able to name the expensive tools used to measure these. They should also be able to produce evidence of previous successful implementation of networks with 10, 40 or 100 Gb/s. Modular patch panels and MTP/MPO connections are also an essential feature in a specialist’s portfolio.
Some providers prepare quotes for active network components without giving due consideration to the existing cabling. Our approach is to clarify these things beforehand—ideally together with the electrician and the customer. In an average year, we design many new networks while our customers’ IT departments only tackle fundamental design changes or a complete network overhaul only once every five to ten years.
Trying to patch something after the fact is much more hassle than good planning from the beginning. Together with our specialised partners, we help our customers design highly customised solutions that really fit.
Machine-to-machine communication and a multitude of sensors can require extensive cabling, and Ethernet cables aren’t always the best choice. KNX/EIB cabling can be used in ring, star and line topologies. Data is distributed through electrical switch cabinets with Ethernet-enabled components mounted on DIN rails. Proximity, temperature and other sensors aren’t exclusive to industrial environments. For instance, hotels, hospitals and office buildings can also benefit from intelligently linking their lighting, smoke detectors, access control, air conditioning and window shading. Trying to patch something after the fact is much more hassle than good planning from the beginning. Together with our specialised partners, we help our customers design highly customised solutions that really fit.
Wireless solutions are often essential for mobile applications. Given the necessary interfaces or upgrade options, connecting desktop equipment to a Wi-Fi network isn’t rocket science, either. In fact, a number of our customers already use Wi-Fi as their primary access method for end devices. This is particularly useful in rented and listed buildings. However, the preferred way to power an access point is still through a PoE (Power over Ethernet)-enabled LAN port. So while Wi-Fi means you can dramatically cut down on cabling, you can’t do away with it altogether. Plus, cables still support higher bandwidths. In a Wi-Fi, users often depend on shared frequencies, and then there’s the matter of interference. And lastly, redundant configurations and failover ensure much better business continuity in a wired network. In particular in industrial environments the choice of connectivity boils down to the question of what sort of latency can be tolerated for specific applications. A need for real-time communication can nip a Wi-Fi network in the bud.
Only when you know the workload during peak times can you determine the optimal setup and invest accordingly.
OFaster speeds can often be achieved without replacing the existing cabling at all. High-speed network components are becoming more and more affordable. With every new generation of transceivers and switches, the price of the previous generation drops. For example, 25 Gb/s is replacing 10 Gb/s as the preferred speed for standard LAN interfaces for servers. 40, 50 and 100 Gb/s are also now available and are being used for top-of-rack and switch-to-switch connectivity not just in large-scale data centres, but also in much more compact data distributors in mid-sized enterprises. And the 200 Gb/s standard is soon to follow. But only if you know the workload during peak times can you determine the optimal setup and invest accordingly. Many customers simply replace all active components in one fell swoop to reach the next speed, without measuring or analysing—with a sound network management system—what their system can already do. Video conferencing and vMotion are just two of the requirements that motivate customers to play it safe when it comes to bandwidth capacity. However, it can indeed be more efficient to simply optimise an existing configuration to effectively unburden the existing equipment without replacing a single component.
Some end devices can still work perfectly well with a bandwidth of 100 Mb/s, so yes. A more likely culprit for bottlenecks is how they connect to distribution switches. If an access switch only supports 1 to 2 Gigabits, then this bandwidth is shared by all end devices. If the same switch is connected to access points, Wi-Fi traffic takes up even more uplink bandwidth. The extra speed that comes with modern wireless standards may well be curbed in the next network rack.
That may well be. But only certain manufacturers produce compatible devices. If you want to avoid costly and complex cabling, go for a top-of-rack design comprised of two switches per rack that redundantly supply it with data connections. If you also want to connect servers with fibre cables, there will only be very limited ports available for each switch. A modular system that occupies a single rack unit could provide a clever solution here. Sometimes it’s more efficient to do away with 10G Base-T in favour of consistent fibre connections and fewer network hardware.
Because the extra speed that comes with modern wireless standards may well be curbed in the next network rack.
Devices with 40 Gigabits and more are generally dismissed in the preliminary stages for reasons of cost, often because there is no need for these kinds of speed yet. But in many switch systems, a 40 Gb port can be split into fully fledged 4× 10 Gb ports. And with a fanout/breakout panel in the server room, you can simply add cables later on with plug-and-play convenience. And you’ll no longer require an electrician to provide splicing. A more expensive 40G switch with 32 ports can thus replace two to three 48-port 10GE switches, which is already worth it in terms of procurement and especially when you factor in service and support. Some manufacturers offer adapters for 100/50/40/10 and 1 Gigabit connectivity plus copper interfaces all rolled into a single device.
For a good picture of the possibilities out there, there’s no way around digging for data sheets and studying them in detail. Not all transceivers are supported by every device. In addition to industry standards, some manufacturers also have their own transceiver types for special requirements. These are not, however, compatible with solutions from other manufacturers, or there may be a gap in service and support.
If you know the cable lengths and types of your passive network environment, you can find useful information about their capabilities in the length restriction chart.
Published on Jun 5, 2018.