< Previousserviced while the rest remain online. As a result, the latest UPSs provide greater reliability, availability and resilience at a lower cost. And while UPSs traditionally relied on lead-acid batteries, these are now being complemented by lithium-ion batteries that are smaller, lighter and better suited for bridging longer periods without power. Add to this the development of supercapacitors, which are not batteries in their own right but can store and rapidly discharge large amounts of electricity, and entirely new features become possible. Smarter grids thanks to UPSs In other words, the role of UPS technology is moving from the protection of critical loads to the protection of the grid at large. The latest generation of UPSs are grid-interactive, meaning they can handle bidirectional rather than just unidirectional energy flows. Instead of simply drawing power, these smart UPSs are also able to feed excess energy back to the grid – without any compromises to their operations. On the one hand this facilitates short and long term peak shaving to reduce the load on the grid. On the other hand it supports the expansion of renewables, which are subject to greater fluctuations than conventional, fossil energy sources. By enabling grid stabilisation, more local power management and advanced control of energy flows, modern UPSs can thus make an important contribution to the energy transition. Further developments will be less about improving the efficiency of the UPS itself, but more about how it can control, store and leverage different energy sources in the interest of sustainability. For instance, if the electricity stored in the UPS of a data centre comes from 100% renewables, then power losses of 3% do not create any additional emissions and have less of an impact on sustainability. Future challenges Increasingly, the focus on life-cycle emissions will shift away from those caused by the operations of the UPS to the sustainability of the manufacturer’s processes. Also known as cradle to gate, this approach emphasises the carbon footprint of a product from the start of production to the moment it is sold. For customers, it is thus becoming ever more important what resources and how many go into making a UPS, including the type and amount of energy consumed during production. With UPSs approaching the limits of what’s possible in terms of efficiency gains, attention is now being paid to other aspects such as the power source of backup generators, which is still frequently diesel fuel or gas, neither of which is ideal. In the discussion around the services that UPSs can perform for the grid, it’s also important not to lose sight of their primary purpose, namely to deliver clean, uninterrupted power to critical equipment. That being said, the UPS of the future is able to do much more than that, by acting as an essential building block in the transition toward net zero. But even when power is available, its poor quality may threaten equipment through voltage surges, sags and transients, particularly in the case of today’s ever more miniaturised and susceptible devices. A UPS safeguards the functioning of these devices by “smoothing” the voltage provided by the grid. Recent developments in UPS technology While the first such devices were analogue and had a capacity of only a few kilowatts, modern UPSs with digital signal processors can handle large volumes of data, which allows for much more advanced functions and control, combined with several megawatts of power. In parallel, UPS efficiency has improved, with power losses reduced from about 20% 30 years ago to 3% today and the use of smaller, transformerless designs that require less raw material. Moreover, the latest models benefit from faster switching, enhanced power electronics and more efficient cooling – all of which are important considerations in the face of rising energy and cooling costs and the drive towards decarbonisation. Together, these advances have enabled modular set-ups that offer easy scalability and maintenance – all that needs to be done to expand capacity is to add another module, and individual modules can even be 2019 2019 2018 2009 2004 1996 1993 1989 1987 1962 1972 Eaton Gigabit Network Card – the first UPS connectivity device to meet the UL 2900-1 cybersecurity standard EnergyAware UPS - The first solution that...etc. enables data centres to contribute to renewable energy and generate revenues from the necessary investments. UPSG Container – Back up power for hours, allows the use of high starting current loads in a rated power grid. First UPS with double-conversion protection and 99% efficiency (ESS mode) up to 1100 kVA Modular transformer-free UPS up to 160 kVA, integrating an unmatched combination of power performance. First UPS with wireless paralleling (HotSync) First UPS with ABM technology to extend battery life First high-frequency transformer-free UPS First UPS with advanced pulse-width modulated (PWM) technology and microprocessor-based diagnostics First UPS with advanced pulse-width modulated (PWM) technology and microprocessor-based diagnostics First all digitally-controlled UPS Technology Innovations through the years 2021 The first release of enhanced ECT (Easy Capacity Test) to test UPS installation including up/downstream switchgear without load bank, reducing cost and CO2. ups technology www.networkseuropemagazine.com 40Make the most of your presence NETWORKS EUROPE magazine is the longest established and industry leading technical journal for the network infrastructure and data centre marketplace. • NETWORKS EUROPE features editorial contributions from worldwide industry figureheads, ensuring that it’s the world’s best publication for information on all aspects of this constantly evolving industry. • Published every other month (x6 per annum), the magazine is produced in digital format, with a magazine viewing link (readable on all major electronic devices) e-mailed directly to subscribers on publication. • The readership consists of 26,000 industry professionals across Europe; with its core circulation covering the UK, Germany, France, Belgium, The Netherlands, Italy and Spain. • The magazine's highly focused editorial content caters exclusively for an informed audience consisting of network infrastructure professionals, including; data centre managers, facilities managers, CIOs, CTOs, ICT directors, consultants and project managers. • Key editorial content areas include; news, legislation and technical information from industry-leading companies and commentators, with detailed case studies, as well as the latest thinking in technology and practices. Advertising Advertising can be in the form of company or product promotion. You can contact our advertising team for details on costs. We accept adverts that are submitted to us in the form of image files saved as high resolution (>300dpi) *.pdf, *.png, *.jpg or *.eps format files. Sponsored content We publish sponsored or branded content in the form of advertorials, case studies, white papers and product/company features. Our advertising team can help with advice and costs. Contact sales@networkseuropemagazine.com for more details. NETWORKS EUROPE The magazine for network and data centre professionals www.networkseuropemagazine.com 41When Power over Ethernet (PoE) was first introduced, it is doubtful that the creators could have imagined how far the technology would be pushed in terms of the amount of power it would supply. Dan Barrera, Global Product Manager, TREND Networks discusses. PoE standardisation When Power over Ethernet (PoE) was first introduced, it is doubtful that the creators could have imagined how far the technology would be pushed in terms of the amount of power it would supply. The very first implementations provided just four watts, enough to power a nightlight. Today, PoE can supply more than 90 watts, enough to power a 75in flat- panel TV. The standardisation of PoE took three steps over 15 years, beginning with IEEE 802.3af in 2003, 802.3at in 2009, and 802.3bt in 2018. Each step increased the amount of power that could be delivered to attached devices by increasing the number of pairs, voltage and current used. The 802.3bt version was the most complicated for several reasons. The amount of power being delivered can create heat in both the cable and in the power supplies, which could lead to signalling errors and shortened service life of the PSE (Power Sourcing Equipment). Throughout the PoE development process, both the ANSI/TIA’s TR42.7 and the ISO/IEC’s SC25/ WG3 engineering committees worked on the standardisation process for the cabling that would be required to support 802.3bt PoE, also known as PoE++. Testing revealed that minor differences in the resistance between the pairs and within a pair of cables caused problems with PoE++ delivery. The problem stems from the way power is applied to the twisted pairs for PoE use. Compared to data signals which are differential and apply opposite voltages to each conductor of a pair, PoE is a common-mode voltage where the same voltage is applied to both conductors in a pair. Two pairs are the positive voltage while the other Dan Barrera, Global Product Manager, TREND Networks what you need to know Resistance unbalance measurements for PoE testing: Figure 1: PoE Wiring specifications PoE testing www.networkseuropemagazine.com 42ance r two pairs are the negative voltage. If the resistance with a pair is different, or the resistance between two pairs differs, the wire acts as a resistive heating element. Think of it as the wires in an old electric heater, just on a much lower power level. In this case, two problems arise; cable heating and PSE heating. If the PSE encounters additional resistance while trying to force high levels of power through the cable, its internal circuitry will heat up. The additional heat may reduce the service life of the equipment, reduce efficiency and increase the costs of operation from increased cooling needs in the data centre to compensate. Secondly, using the electric heater analogy, the cable’s temperature will increase as power is forced through it unbalanced. Fortunately, it’s not going to become glowing hot, but the temperature increase can get high enough to cause signal transmission issues. Cable insertion loss (attenuation) Figure 2: PoE Classifications PoE testing www.networkseuropemagazine.com 43is affected by three factors: length, signal frequency and temperature. In cases where the installed length is near the 90m limit and the cable is installed in an already hot environment, an increase in temperature from PoE use can cause the insertion loss to exceed levels where packet errors occur. How can we detect these problems? The DC Resistance Unbalance (DCRU) is the measurement that is performed on the cabling after installation to test for differences in DC resistance. It is performed by cable certifiers in addition to the radio frequency (RF) measurements to test cable performance such as NEXT, insertion loss and return loss. DCRU measures pair unbalance and pair-to-pair (P2P) unbalance. Testing DCRU ensures that the installed cabling meets the stringent unbalance requirements of the TIA and ISO standards. The requirements are a maximum pair unbalance of 7% or 0.20 ohms max and P2P unbalance 7% or 0.20 ohms max. Think about measuring pair unbalance as comparing the resistance of two separate ohm meters while measuring the conductors in a pair. In the diagram (Figure 3) both positive leads are attached to the same side of the cable, each on a different conductor; and on the opposite end, the negative leads are attached to the corresponding conductors. Each meter is measuring the resistance of its conductor and the two values are compared. Pair unbalance is measured for all four pairs. According to the specification, they must be within 7% of each other or differ by no more than 0.20 ohms. Measuring P2P unbalance is a comparison of the parallel resistance of the pairs compared to one another. In the diagram, the green and orange pairs are compared for demonstration. For an actual DCRU test, all four pairs are measured and the difference between the lowest and highest measured value cannot exceed 7% or 0.20 ohms. Causes of resistance unbalance Resistance unbalance can be inherent in the cable or can be a result of cable termination. Pair unbalance, the difference between conductors in the same pair, is unlikely in the cable. It is usually going to be the result of a poor termination of one of the conductors. Pair-to-pair unbalance, the difference between pairs, can come from the cable when lengths exceed 90m. While the TIA limits link to 90m and channels to 100m, the ISO standards do not have a maximum length limit. ISO 11801-1/IEC 61935-1 only require that the length be reported and that the RF test parameters pass. Therefore, a link can be long enough where the different twist rates can lead one pair to be significantly longer than another which can cause the P2P unbalance to exceed specifications. Termination of the cable to the connector is another common cause of resistance unbalance. Poor punch-downs can cause high-resistance connections that pass RF certification tests. This is because the RF energy travels on the surface of the conductors and if a conductor is barely touching the IDC (insulation displacement connector) contacts, enough energy will pass through the connection so that tests like insertion loss and return loss will be unaffected. The same connection will have a high DC resistance measurement causing a resistance unbalance. Is DCRU testing mandatory? The latest TIA, ISO and EN/CENELEC cabling standards do not require field testing DCRU. It is entirely optional. In 2020 the TIA produced an addendum to the 568.2-D, the 568.2-D-2 for power delivery that defines the DCRU requirements for all channels from category 5e through 8. Still, it is up to designers and specifiers to understand the importance of DCRU in how it impacts the ability of a cabling system to support PoE++. Best practice is to require DCRU testing in addition to the usual certification testing. While any cabling from category 5e and up should pass DCRU testing, some companies are marketing cable specifically for PoE++ applications. This usually involves using a larger wire gauge (22 AWG typically) on cat 5e or cat 6 cables that might usually be constructed of 24 AWG conductors. The larger wire will reduce loop resistance to ensure 90+ watts can be delivered on 90m links. Testing DCRU in the field While all cable certifiers can measure DC loop resistance (a requirement when they were designed), not all cable certifiers currently deployed in the field, or even some being sold today, can measure DCRU. Because it is a relatively new test, only certifiers designed within the last few years can measure DCRU. This means that you’ll need a fairly new certifier to be able to test DCRU. Loop resistance should also be measured when certifying for PoE support because it is an important parameter in ensuring the full amount of power can be delivered to the PoE device. Since none of the certification test standards require DCRU testing, your DCRU enabled Figure 3: Pair unbalance measurement configuration Figure 4: Pair-to-pair unbalance measurement configuration Figure 5: Selecting a +PoE test standard PoE testing www.networkseuropemagazine.com 44certifier will have a setting to add DCRU to the regular Autotest for the category/class being certified. In figure 5, the certifier has an option for Cat5e + PoE which adds DC loop resistance and DCRU to the required TIA tests to certify a link for cat5e performance. The series of screens shows the results from the loop resistance and DCRU tests. Loop resistance is the total resistance of both conductors in a pair. The limit is a fixed value regardless of cable length. Pair-to-pair resistance (figure 7) is the parallel resistance of each pair compared with the other three pairs in the cable. There are six combinations that need to be compared when measuring P2P resistance. In this example, the worst case is between pairs 12-36 with a value of 0.15 ohms. Pair resistance (figure 8) is the difference between the conductors in each pair. In this example, all four pairs pass with pair 1-2 having the highest unbalance of 0.24 ohms. In this example pair 7-8 (in figure 9) failed the unbalance test with a value of 0.92 ohms which is 0.46 ohms over the limit. The first step in troubleshooting should be to re-terminate the connector. The Autotest results for the same cable (figure 10) show that NEXT, insertion loss, and return loss all pass, meaning that this cable will support data transmission. The poor termination would go unnoticed if the DCRU test was not enabled. Summary As PoE becomes a necessary part of every network, field testing should be performed to ensure the materials and the installation will support high power PoE without generating excess heat. The result can cause data transmission issues, shorten the service life of the power supplies, and reduce overall electrical efficiency. Even though DCRU testing is not required by the cabling standards, a best practice is to test it anyway. This will ensure the cabling and the installation workmanship is up to the task of supporting PoE to its fullest capability. Can your cable certifier be upgraded to support DCRU testing? Some manufacturers such as TREND Networks offer DCRU adapters for previous generation certifiers that extend their useful life by allowing them to test DCRU with a permanent link and channel adapters that have the precision resistance measurement circuitry built in. The new adapters and a software update allow their LanTEK III certifier to measure DCRU along with the LanTEK IV which has integrated DCRU measurement capability. Check with your certifier manufacturer to see what options are available for DCRU testing as it becomes a common requirement and PoE++ capability is increasingly a necessity for new cabling installations. Figure 6: DC loop resistance Figure 7: Pair-to-pair resistance, Pass Figure 10: Autotest result Figure 8: Pair resistance, Pass Figure 9: Pair resistance, Fail PoE testing www.networkseuropemagazine.com 45How to fast-track the data centre specification process specification process www.networkseuropemagazine.com 46Key drivers of data centre modernisation projects include the need for greater flexibility, increased resilience or reduced risk, and lower operating costs. However, various legacy deployments lack standardisation and need better visibility into the operational environment before such projects begin. For those that are suffering from high energy costs, stranded capacity, or hotspots, it can often require a complete overhaul of the infrastructure. For many data centre managers, beginning with the end in mind is the simplest way to fast-track the deployment process and by utilising a data-driven approach, end-users can meet business objectives without risking outages or downtime. By first assessing the existing infrastructure environment, for example, and utilising computational fluid dynamics (CFD), Andy Connor, Channel Director EMEA, Subzero Engineering Todays demands for data centre performance, reliability, and sustainability are pushing the boundaries of physical infrastructure designs and with many businesses undertaking complex modernisation strategies, the need to simplify the design and specification process has become ever more pressing. www.networkseuropemagazine.com 47 specification processoperators can identify key areas of concern within their systems and take steps to reduce hotspots, manage poor airflows and address inefficient rack configurations. Once the existing infrastructure has been assessed, data centre optimisation initiatives can begin, and end-users may choose to completely replace a legacy system if the costs of modernisation to an old or outdated system outweigh the long-term lifecycle benefits. Instead, retrofit projects can allow data centre operators to re-design the whitespace and improve power and cooling efficiencies within a new containerised architecture. An approach such as this also offers the ability to increase rack densities, optimise performance, and reduce operating expenditure (OpEx). Key considerations to specify for your infrastructure environment The truth, however, is that every data centre is unique in its design and infrastructure capabilities. Therefore, a key part of any modernisation program is the specification process, ensuring that any new mission- critical environment will meet the desired standard from the outset. Here, taking a standardised approach to the data centre design and combining it with a detailed structural analysis can pay huge dividends, helping any external partners or consultants to understand the technological choices, and ensure any demanding timescales can be adhered to. Key areas to document will include the size of the system, its power requirements and its cooling configuration - primarily whether or not the data centre will utilise a hot or cold aisle cooling architecture - and the floor type, whether raised or slab, on which it is to be deployed. Specifying containment is also crucial, especially if the project is centred around improving energy efficiency, reducing Power Usage Effectiveness (PUE) and CO2 emissions, or lowering OpEx. The benefits of containment systems include improved airflows, maximised whitespace and reduced hotspots within the mission-critical environment. Further, by optimising a legacy facility with a containment system, end-users can often achieve an average PUE reduction of 0.4, which will have a significant and beneficial impact on operating costs. Other key components to specify will include the specification process www.networkseuropemagazine.com 48number of internal and external support arm tiers, their length, any blanking panels for the racks, and ultimately, any door configurations to accompany the containment system. Gartner, for example, predicts that by 2025, data centres deploying speciality cooling and density techniques will see 20% to 40% reductions in operating costs, so optimising your design and cooling strategy is essential from the beginning of the project. Build faster and better The system’s structure, ceiling height and type are also important factors to specify, and the width of aisles or any obstructions such as columns within data halls can also influence system design and configuration. Rack size, height and colour preference must be documented from the outset, and any customised height or colour configurations can have an increased lead time. It’s essential, therefore, for external consultants to set clear expectations and communicate how any delays to the supply chain may impact deployment times. Many traditional methods for supporting data centre infrastructure including containment, power distribution and cable routing can be costly and time-consuming if not properly designed or specified. They often require structural ceilings, underfloor pathways and a building that can support the entire weight of the environment. Specifying the load of the system, therefore, is crucial and will influence how the system is designed and configured. Here, pre-configured data centre systems can provide a simplified and quick-to-deploy architecture that includes infrastructure conveyancing, support and containment. Essential structures such as this can allow mission-critical facilities to be built to scale and in demanding timescales, both quickly and efficiently, including all power and cooling components. By working closely with customers throughout the design and specification phases, through the manufacturing process and throughout installation and commissioning, infrastructure manufacturers can ensure that any data centre modernisation project meets its business objectives from the outset. specification process www.networkseuropemagazine.com 49Next >