To Solve Data Centers’ Downtime and Efficiency Problems, Go to the Source
About one-third of reported outages cost $250,000, and many exceed $1 million. There is also the global movement toward reduced carbon emissions, causing data centers to develop innovative cooling solutions and other remedies to offset their consumption. What’s more, emerging technologies — such as Artificial Intelligence (AI), Machine Learning (ML) and Internet of Things (IoT) — will demand ever more processing power out of data centers.
Some interesting innovations have developed from these pressures, including using data centers to heat swimming pools and homes and even powering farms. On-site alternative energy generation and direct connections to nuclear power plants are also being brought online. These developments, however, do not address the issues of downtime and power usage at the source. All the aforementioned pressures drive a need for more innovative and effective ways to monitor power quality along a data center’s entire power chain. The need for reliable, intelligent rack power distribution, monitoring and control has never been greater.
Data center operators cannot get all the information they need from today’s typical intelligent rack PDUs. The true key to minimizing downtime, optimizing electrical usage and meeting sustainability goals is a next-generation intelligent rack PDU with industry-leading visibility, reporting and alerting for power metrics and events at the cabinet, as well as best-in-class flexibility to meet and anticipate future requirements.
Load Management
Power quality assessments are typically one-off events when a data center is brought online or after a major server overhaul. These assessments fail to take into account how power quality issues affect a facility on an ongoing basis or how normal operations might cause power quality problems.
A rack at a data center is likely going to be powering network switches, storage devices and different types of servers ― including blade servers, “pizza box” servers and others in varying styles and configurations. This equipment all runs on switch-mode power supplies, meaning the power supply fluctuates depending on the devices’ needs. Switch-mode power supply units, variable speed drives, computers and other equipment carry non-linear loads, which can cause harmonic distortion. This can lead to many problems, including overheating cables, false device trips, energy loss and equipment failure.
A typical intelligent rack PDU is incapable of providing the visibility required to monitor and manage these types of issues at the rack and device level. This lack of visibility also hampers responses to events such as circuit breaker trips. The time-consuming task of identifying a device that caused a trip can be eliminated with a truly intelligent rack PDU with advanced forensics showing data at the device level. It should also support capacity planning, workload or environmental optimization, failover planning and uptime/troubleshooting (both proactive and reactive) initiatives.
Without a high level visibility at the rack and outlet level, it becomes difficult to optimize efficiency and infrastructure. Power usage effectiveness (PUE) is the most common way data centers measure their efficiency. It is the total power consumed by a data center divided by power used by IT equipment. While a helpful gauge, it does not show when and where power is being used inefficiently. It is important to ensure power is actually being used to support mission-critical work at all times.
Barriers to Efficiency
The lack of visibility also limits operational efficiency and infrastructure scaling. Being unable to identify stranded capacity and other issues make it difficult to maximize space and usage effectively, hindering the ability to deploy high-density IT infrastructure. Another hurdle to maximizing density is a limited number of outlets — or the wrong kind — at the rack level. An intelligent rack PDU should provide flexibility in outlet provision to support high-density deployment.
A high-visibility power quality monitoring and management solution can also identify servers that draw significant amounts of power even when they’re idle. Additionally, it can give insight into outlet issues and concurrent events. These causes of costly downtime could all be addressed with metering solutions for monitoring and addressing power quality issues at the rack (and device) level.
The typical intelligent rack PDUs present further challenges to data center operators in that they’re not seamlessly compatible with building management systems (BMS) and data center infrastructure management (DCIM) solutions already in place. This requires data center operators to procure and implement additional tools and features to maximize uptime and efficiency. The ideal solution should connect to existing DCIM or BMS solutions, whether they were built in house or provided by a third party. This simplifies how facility managers collect and transmit the valuable data they can capture from the outlet, rack and environment around the rack PDU.
Back to the Source
If the goal is maximizing uptime and efficiency, the work starts at the source. That’s why an intelligent rack PDU with visibility at the rack and device level is so critical to achieving data centers’ goals and ensuring they withstand the pressures they are facing today. The right solution will also prepare data center operators for the demands of the future as technology advances.
Raritan, a brand of Legrand, has launched the forward-thinking PX4 Intelligent Rack PDU, which provides data centers with visibility, reporting, security, and alerting, helping to drive operational excellence.
Prabhakar Muthuswamy is Sr. Product Manager at Raritan, a brand of Legrand. Discover more about the PX4 PDU.
