Event Logging: An Essential Tool for Troubleshooting Power Problems

Power events such as transients can last only microseconds while other power disturbances (e.g., blackouts, brownouts, sags, etc.) may persist for only milliseconds. Both are much shorter than the average time to blink your eye. Intermittent power disturbances caused by harmonic distortions often occur randomly making them hard to troubleshoot even moments after the power disturbance occurs. [Read more…]

Why Hot-Swap Capability is Key to Zero Network Downtime

In the past, UPS repairs or maintenance were often a big headache for IT departments with UPS Systems that were non-modular and hardwired making them time-consuming to repair. An off-site technician was usually required to diagnose the problem, often causing an undesirable period of network downtime. But today, when high availability, resiliency and serviceability are critical requirements for network operations, UPS manufacturers have addressed the problem of downtime by engineering Hot-Swappable UPS Systems. [Read more…]

“Selling” 3-Phase Power to Management

You don’t need to know much about 3-phase power to decide whether you need it for your application. If you decide that 3-phase power is the best choice for your needs, there are some attributes of 3-phase power that can help you sell the idea to the executive suite if you get push back over doing something new and different. [Read more…]

Improving Power Usage Effectiveness (PUE)

What is Power Usage Effectiveness?
Power Usage Effectiveness (PUE) is the ratio of total energy employed at a data center or network facility vs. the total energy used strictly by IT equipment. Essentially, it measures how efficiently power is utilized in the data center. This metric is ideally used as the basis to make the data center more efficient.

How to Measure PUE
PUE = Total Facility Energy / IT Equipment Energy

Given the equation above, we can infer that two values are needed to determine PUE: Total Facility Energy and IT Equipment Energy. To calculate each, you must make a comprehensive list of all energy-consuming equipment and then categorize each item as part of either a.) the IT load or b.) the physical infrastructure. Once your list is categorized properly, you can use the two values to solve for your data center’s power usage effectiveness.

Improving PUE
An ideal PUE is 1, or 100% efficiency. The lower (or closer to 1) the PUE, the more efficient the data center. The more efficient the data center, the lower the total cost of operation (TCO).

There are a number of ways to improve PUE, but you should start by identifying the low-hanging fruit. For example, cooling is often the biggest offender in low-efficiency data centers. You can reduce the power required for cooling by taking some simple steps to increase airflow:

1. Organize cables
Bulky cabling can obstruct airflow, especially if it’s disorganized and tangled. Using the right type of rack with the appropriate cable management will allow air to flow freely and cool equipment properly.

2. Install Blanking Panels
Blanking Panels can help force cold air through servers and other equipment. They help define the path of cold air so that less of it is wasted and more of it is distributed directly to equipment.

3. Isolate and Remove Hot Air
Make sure that hot air has a way out. If the exit path isn’t well-defined or isolated, there is greater risk of the heated air recirculating into the data center. Since heat rises, the best exit path is through a raised ceiling panel or roof-mounted thermal duct and onward to the facility’s built-in exhaust system.

Once you’ve taken these basic steps, you may want to look at other ways to improve PUE. As data centers are complex environments with a lot of moving parts, there are numerous places to look for inefficiencies. Consult with our Technical Support Team to determine other methods for increasing efficiency.

Why Monitor Power Consumption in a Server Rack?

If rack power consumption could be easily estimated and remained constant, there would be no need for on-going monitoring. However, server workload and rack configuration changes cause power consumption to fluctuate over time and have the potential to overload the circuit, causing equipment to shut down.

The problem
Power consumption in a server rack is not constant. In fact, it can more than double when servers are performing processor- or disk-intensive operations such as copying files. Power consumption can also vary by as much as 20% in servers running Windows Server, which automatically reduces the frequency of the server’s processors to save power when the machine is under low load. Variations in the performance characteristics of servers  makes it difficult for network engineers to estimate power requirements for a server rack and ensure that an overload condition does not occur.

The solution
To ensure that rack power consumption never exceeds the capacity of the circuit (protected mains, in-rack UPS or branch PDU), a network engineer could design-in a large margin of error but this would greatly increase both capital and on-going costs. A better solution is to continuously measure the power consumption of each rack in real time using a PDU capable of remote monitoring.

A Monitored PDU supports real-time local and remote monitoring of the power consumed by connected equipment.  With optional accessories, temperature, humidity, and rack security can also be monitored. If power consumption exceeds a user-configured threshold, an email or similar alert warns network engineers of potential overload conditions.