42:24
PUE Essentials: Your Guide for Measuring & Improving Data Center Energy Efficiency
SOCOMEC, INC. - USA
Overview
This video explains the concept of Power Usage Effectiveness (PUE) as a key metric for data center energy efficiency. It details different types of data centers (hyperscale, enterprise, colocation) and emphasizes the importance of power monitoring for measuring and improving PUE. The presentation covers how power monitoring enables capacity planning, sub-billing in colocation facilities, and proactive maintenance by identifying energy consumption patterns and power quality issues. It also outlines specific data center power distribution systems (RPP and busway) and where to implement metering for accurate PUE calculation, concluding with best practices for effective power monitoring.
How was this?
Save this permanently with flashcards, quizzes, and AI chat
Chapters
- The video introduces Power Usage Effectiveness (PUE) as a critical metric for data center energy efficiency.
- It outlines different types of data centers: hyperscale (large, single-company owned), enterprise (company-specific IT needs), and multi-tenant/colocation (rented space).
- Soch, a 102-year-old company, offers North American solutions including metering devices (CCS) and power distribution systems (PowerSmith).
Understanding the different data center types and the solutions available helps contextualize the importance of energy efficiency and the role of monitoring tools.
Examples of data center types include Google/Amazon (hyperscale), Bank of America/Boeing (enterprise), and Digital Realty/Equinix (colocation).
- PUE is calculated as total data center energy consumed divided by the energy used by IT equipment (servers).
- The ideal PUE is close to 1, meaning almost all energy powers IT equipment with minimal waste.
- Power monitoring is crucial for identifying energy consumption hotspots, enabling efficiency improvements, and supporting capacity planning.
- Monitoring also supports economic factors like reducing energy bills and enables accurate sub-billing in colocation environments.
Accurate PUE measurement and understanding energy usage patterns are essential for reducing operational costs, environmental impact, and ensuring system reliability.
The average PUE has decreased from 1.6 before 2020 to 1.15 in 2023, showing progress in efficiency, with the goal being to approach 1.
- Power monitoring allows data center operators to assess if adding more servers will overload the system, aiding in capacity planning.
- It helps predict peak energy usage times.
- Monitoring power quality, including harmonics and phase balancing, is vital for critical 'emergency loads' like data centers to prevent downtime.
- Understanding incoming power quality from the utility provider helps in managing issues like voltage sags.
Proactive monitoring of capacity and power quality prevents costly downtime, ensures system stability, and allows for informed decisions about infrastructure upgrades.
Identifying a PDU breaker that is not performing well allows for scheduled maintenance before it fails, preventing downtime.
- Two main distribution systems in North America are Remote Power Panels (RPP) and Busway.
- RPP systems use PDUs to feed local panels near server racks, with metering points at main incoming power, PDUs, and RPPs.
- Busway systems connect PDUs directly to busways, with tap-offs for server racks, and metering points at main incoming power, PDUs, and each tap-off.
- Accurate PUE calculation requires metering at multiple levels, ideally as close to the IT equipment as possible (PUE3).
Knowing where to implement power monitoring within different distribution architectures is key to gathering the granular data needed for precise PUE calculations and system analysis.
For RPP systems, metering can be implemented at the main incoming power, the PDU, and the RPP itself to capture energy flow.
- PUE calculation involves different levels: PUE1 (before PDUs), PUE2 (before distribution to racks), and PUE3 (closest to IT equipment).
- PUE3 is the most critical for accurate PUE calculation, providing detailed insights into server-level performance and load balancing.
- Soch offers solutions like gateways, voltage/current modules, and power quality meters (e.g., A200) for PDUs.
- They also provide metering for busway tap-offs and RPPs using modular systems like Digiware BCM for branch circuit monitoring.
Understanding the different PUE calculation levels and how Soch's modular metering solutions can be applied at each stage ensures comprehensive data collection for optimal efficiency.
Soch's Digiware system with current modules and sensors can be used to meter breakers downstream on a PDU, feeding into RPP boards.
- Implement permanent, continuous monitoring rather than temporary spot checks for reliable PUE data.
- Ensure granular metering (PUE3) as close to the IT load as possible for the most accurate readings.
- Choose scalable and versatile systems that can adapt to future data center growth and measure at various points.
- Prioritize accuracy, especially at low load conditions, to ensure reliable data across all operational states.
Adhering to these best practices ensures that power monitoring systems provide the accurate, continuous, and granular data necessary to achieve and maintain optimal data center energy efficiency.
Soch's metering solutions, using current transformers (CTs), offer accuracy of 1% even at 1% load, outperforming revenue-grade metering requirements.
Key takeaways
- PUE is the primary metric for data center energy efficiency, calculated by dividing total facility energy by IT equipment energy.
- Comprehensive power monitoring across multiple points (main, PDU, RPP, busway tap-offs) is essential for accurate PUE calculation.
- Metering as close to the IT load (PUE3) provides the most granular and actionable data for efficiency improvements.
- Power monitoring supports critical functions like capacity planning, preventing overloads, and enabling accurate sub-billing in colocation centers.
- Understanding and monitoring power quality (harmonics, voltage sags) is crucial for preventing downtime in critical data center operations.
- Scalable, accurate, and permanent monitoring solutions are necessary for long-term energy efficiency management.
- Soch offers modular and adaptable metering solutions that can be integrated into new or existing data center infrastructure for both AC and DC systems.
Key terms
Power Usage Effectiveness (PUE)Hyperscale Data CenterEnterprise Data CenterColocation Data CenterPower MonitoringIT EquipmentRemote Power Panel (RPP)Busway DistributionPower Distribution Unit (PDU)Capacity PlanningPower QualityHarmonicsVoltage SagGranularityModular System
Test your understanding
- What is the formula for calculating PUE, and what does a PUE value close to 1 signify?
- Why is continuous, granular power monitoring at the PUE3 level considered crucial for data center efficiency?
- How does power monitoring contribute to capacity planning and preventing system overloads when adding new IT equipment?
- What are the key differences between RPP and Busway power distribution systems in terms of metering implementation?
- How can power quality monitoring, such as detecting voltage sags, help prevent downtime in a data center?