Search for: All records

Proponents of AC-powered data centers have implicitly assumed that the electrical load presented to all three phases of an AC data center are balanced. To assure this, servers are connected to the AC power phases to present identical loads, assuming an uniform expected utilization level for each server. We present an experimental study that demonstrates that with the inevitable temporal changes in server workloads or with dynamic sever capacity management based on known daily load patterns, balanced electrical loading across all power phases cannot be maintained. Such imbalances introduce a reactive power component that represents an effective power loss and brings down the overall energy efficiency of the data center, thereby resulting in a handicap against DC-powered data centers where such a loss is absent. 

In recent years, various airflow containment systems have been deployed in data centers to improve the cooling efficiency by minimizing the mixing of hot and cold air streams. The goal of this study is the experimental investigation of passive and active hot aisle containment (HAC) systems. Also investigated, will be the dynamic interaction between HAC and information technology equipment (ITE). In addition, various provisioning levels of HAC are studied. In this study, a chimney exhaust rack (CER) is considered as the HAC system. The rack is populated by 22 commercial 2-RU servers and one network switch. Four scenarios with and without the presence of cold and hot aisle containments are investigated and compared. The transient pressure build-up inside the rack, servers' fan speed, inlet air temperatures (IAT), IT power consumption, and CPU temperatures are monitored and operating data recorded. In addition, IAT of selected servers is measured using external temperature sensors and compared with data available via the Intelligent Platform Management Interface (IPMI). To the best of authors' knowledge, this is the first experimental study in which a HAC system is analyzed using commercial ITE in a white space. It is observed that presence of backpressure can lead to a false high IPMI IAT reading. Consequently, a cascade rise in servers' fan speed is observed, which increases the backpressure and worsen the situation. As a result, the thermal performance of ITE and power consumption of the rack are affected. Furthermore, it is shown that the backpressure can affect the accuracy of common data center efficiency metrics. 

During the lifespan of a data center, power outages and blower cooling failures are common occurrences. Given that data centers have a vital role in modern life, it is especially important to understand these failures and their effects. A previous study [16] showed that cold aisle containment might have a negative impact on IT equipment uptime during a blower failure. This new study further analyzed the impact of containment on IT equipment uptime during a CRAH blower failure. It also compared the IT equipment performance both with and without a pressure relief mechanism implemented in the containment system. The results show that the effect of implementing pressure relief in containment solution on the IT equipment performance and response could vary and depend on the server's airflow, generation and hence types of servers deployed in cold aisle enclosure. The results also showed that when compared to the discrete sensors, the IPMI inlet temperature sensors underestimate the Ride Through Time (RTT) by 32%. This means that the RTT calculations based on the IPMI inlet sensors may be inaccurate due to variations in the sensor readings; as they exist today; in these servers. as discussed in a previous study [26]. Additionally, it was shown that all Dell PowerEdge 2950 servers have a similar IPMI inlet temperature reading, regardless of mounting location. As external system resistance increases during cooling failure, the servers exhibit internal recirculation through their weaker power supply fans, which is reflected in the high IPMI inlet temperature readings. For this server specifically, a pressure relief mechanism reduces the external resistance, thereby eliminating internal recirculation and resulting in lower IPMI inlet temperature readings. This in turn translates to a lower RTT. However, pressure relief showed conflicting results where the discrete sensors showed an increase in inlet temperature when pressure relief was introduced, thereby reducing the RTT. The CPU temperatures conformed with the discrete sensor data, indicating that containment helped increase the RTT of the servers during failure.