By Marc Cram, Director of Sales at Server Technology
Density in the data center refers to the electric power consumed per square foot of floor space (or unit volume of space), number of servers and the cooling systems load. A number of factors in today’s market are driving the data center operator to increase the density and the operational loading of the compute infrastructure.
Among these factors are the desire to reduce both capital and operational expenditures, improve energy efficiency (lower PUE), avoidance of new build-out, reduction of latency times and better management of network traffic flow.
Historically, the aggregate power load of the data center cabinet has ranged from 1-5 kW for a 42U rack. A full rack of single (or dual) corded 1U servers in the aptly nicknamed “pizza box” form factor were the original drivers for a high density (scale out) rack and high outlet count PDU. As the equipment manufacturers adopted a number of new form factors such as 2U, 4U, 5U and 10U enclosures incorporating multiple server chips and large banks of both storage and memory, the demands placed on the power supplies within those servers went up, requiring higher power delivery per cord and outlet, driving the widespread adoption of C19 outlets over the previously dominant C13 outlets.
In the 2018 IHS report “IT racks and enclosures and rack power distribution unit market trackers”, Maggie Shillington of IHS says that “in 2013, rack PDUs over 10kW accounted for 18% of global rack PDU revenue and grew to account for 24.2% by 2017.” The report goes on to state that “While low-power (<5kW) PDUs still accounted for the largest share of the global market in units shipped in 2017, they are the second largest category in terms of revenue.” IHS projects that medium- (5-10kW) and high-power (>10kW) PDUs will continue to grow at faster rates than low-power PDUs over the forecast period in terms of both revenue and units.
There are numerous challenges for both the facilities and IT teams of a data center that are associated with implementing a power-dense data center. Amongst these are cost, heat, load balancing, environmental and power monitoring, increased network traffic and configuration management of systems.
There are companies and industry organizations devoted to improving the state of the art of data center design and power utilization effectiveness. ASHRAE, IEEE, The Green Grid and Emerge Alliance are among those organizations continuously seeking to foster new ideas and new technologies leading to more energy efficient data centers. Most studies and papers from these groups have concluded that minimizing the number of power conversions and operating power supplies at the most efficient (highest) input voltage possible will result in significant savings on operational expenses (OPEX) for the data center. Modern green field data centers are using 480V/277VAC, 415V/240VAC, and 380VDC as typical power going to their racks, with amperages ranging from 16A all the way up to 100A. A 1200mmx800mm cabinet fed with 415V/63A power has a usable capacity of 45.4 kW in a footprint just over 10.3 sq. ft. This configuration represents the practical upper limit of density for most data centers today. This is a typical configuration for many modular data centers that are readily available. Most applications that require higher density are running very specialized applications and are supported by liquid cooling technology.
By making use of the available power and configuration management tools provided by the various server manufacturers, the data center designer and operator can come to a consensus on the best approach to achieving the power and compute density needed for their application. These tools are readily available from manufacturers such as HPE, IBM, and Dell EMC, and they can give the best idea of what level of compute density can be achieved for a given budget.
Another category of software tool that is available to support critical compute loads running in dense data centers is Data Center Infrastructure Management (DCIM). Typically incorporating support for SNMP, most DCIM tools provide a way of measuring power consumption, assessing available power capacity, profiling thermal conditions within the datacenter, exercising closed-loop control of HVAC systems, providing some degree of IT asset tracking, and determining the status of the various elements of UPS that may be distributed throughout the datacenter. DCIM tools help provide a “single pane of glass” for knowing what is going on in the data center. Depending on the DCIM tool selected, it may also provide hooks into any of a number of configuration management and virtualization management packages that enable the dense data center to rapidly configure new servers or move compute loads around to available underutilized resources within the data center.
Computational fluid dynamics software tools are used to determine how the cooling infrastructure is performing relative to the heat loads that are distributed within the data center. CFD will help ensure that hot aisle / cold aisle containment systems are performing optimally while identifying opportunities to redeploy assets to make the best use of available cooling capacity.
Operating the data center at the upper limits of the allowable inlet air temperature range helps achieve lower cooling costs and higher PUE, while requiring the temperature monitoring and control systems to operate reliably and respond quickly to changing conditions. The potential for thermal runaway is exacerbated in this case. The ability for modern PDUs to monitor both power consumption and temperature within the cabinet provides a “last line of defense” in the high temperature environment of the exhaust (hot aisle) side of the datacenter rack. The alerting, alarming, and reporting functions of modern PDUs provide critical information to HVAC systems, DCIM tools, and the IT and Facilities personnel responsible for overseeing the successful operation of the data center. The more individual loads within the racks, the more critical it becomes for the PDU to have granular measurement and reporting capabilities along with supporting the targeted hot aisle temperature.
Placing many servers in a single cabinet plugged into a 3-phase power source necessitates careful planning and monitoring as well. The data center can minimize wasted power and maximize power delivery to the IT gear when the loads on all three phases are as close to equal as possible. This requires the servers and the PDUs to report power consumption and provide an aggregate figure of consumption by phase in order to give the datacenter operator a chance at having a maximally efficient data center.
Utilizing the alternating phase HDOT modules minimizes the potential for accidentally unbalanced loads across the phases. Providing power to a dense heterogeneous computing environment requires highly integrated and extremely reliable power distribution. High Density Outlet Technology (HDOT) Cx from Server Technology provides both. The PRO2 HDOT Cx is available in a “build your own” PDU that allows the user to configure a PDU meeting their exact specification. By selecting the appropriate input power (voltage, amperage, and phases), input cord, PDU orientation, outlet mix, and connectivity, the data center designer and IT specifier can get a PDU that is custom-tailored to their high compute density application. HDOT Cx modules are available in our alternating phase configuration and feature our unique 2-in-1 outlet design that supports both C14 and C20 plugs, thus future- proofing the data center rack by allowing the PDU to be re-used and re-configured more easily than any other data center PDU on the market today.
The PRO2 HDOT Cx delivers on the promise of modern PDU design capabilities. With its modular construction and its completely re-thought outlet layout, PRO2 HDOT Cx provides unique, 2-in-1 outlets that act as both locking C13 and C19 power outlets, all in a standard form factor that fits most data center cabinets. At just 2.2” wide, 2.5” deep, and 70” tall, the PRO2 HDOT Cx PDU is suitable for all server racks 42U and taller. PRO2 HDOT Cx is also capable of operating at full power load in a 60o C environment, allowing the datacenter to run with a warm ambient temperature. The benefits of the alternating phase modules are numerous – they simplify load balancing while minimizing cord lengths, and they make it easier to identify which asset is plugged into each outlet of the PDU.
The Smart, Smart POPS, Switched, and Switched POPS versions of PRO2 HDOT Cx allow for remote monitoring of power and temperature, provide alerting and reporting, thus enabling the datacenter manager to make the most use of the available datacenter compute infrastructure and available power. PRO2 Switched HDOT Cx provides individual outlet control, and POPS delivers power consumption data from the individual outlet. The PRO2 firmware features an open MIB and OID tree information that is freely available for download from the Server Technology website, and PRO2 firmware is supported by more DCIM tools than other PDUs on the market.
With the best initial quality and longest mean time to failure (MTTF), Server Technology products provide uncompromising reliability and value for the data center. Server Technology holds many of the key patents for providing power in a zero-U PDU configuration compatible with today’s racks and dense compute infrastructure requirements.
Our HDOT-based products have been recognized in the data center industry for being the global leader of innovative, intelligent power distribution products. Our PDUs offer solutions for data centers, telecommunications operations and branch offices. Over 60,000 customers around the world rely on Server Technology’s power distribution units (PDUs) and power management and measurement solutions.
About the Author:
Marc Cram is Director of Sales for Server Technology (@ServerTechInc), a brand of Legrand (@Legrand). Marc is driven by a passion to deliver a positive power experience for the data center owner / operator. Marc brings engineering, production, purchasing, marketing, sales, and quality expertise from the automotive, PC, semiconductor and data center industries together to give STI customers an unequalled level of support and guidance through the journey of PDU product definition, selection, and implementation. Marc earned a BSEE from Rice University and has over 30 years of experience in the field of electronics.
