Monitoring systems play a key role in data center infrastructure management (DCIM) to help meet the demand for constant, uninterrupted uptime and efficient operation. They are also an affordable way to monitor assets and temperature to better control energy consumption and spending.
Whether you have a small server room or a large data center, a remote monitoring system can alert personnel immediately when a change in environmental conditions occurs, allowing them to take fast, corrective action. These systems also give operators real-time status of all monitored conditions on demand. Monitoring systems make seeing multiple sensor points easy and provide immediate and historical insight to help operators avoid costly downtime.
Choosing the right monitoring system and components depends on the specific requirements of the data center or computer room. Here are some key conditions to monitor and important factors to consider when investing in a data center monitoring system:
Temperature monitoring plays an important role in DCIM. It allows personnel to identify cold and hot areas so they can raise room temperature where necessary to improve energy consumption and cool areas that might be putting equipment in danger. Monitoring temperature efficiently depends on the size of the room and the concentration of heat sources.
Rising temperature can indicate various problems, like a failing computer room air conditioner (CRAC), computer room air handler (CRAH) or chiller unit, a leaking coolant line, lost utility power, or restricted air circulation. Many CRAC/CRAH systems have built-in alarm fault panels that can be connected to your monitoring system to alert you the moment your cooling system detects an internal problem.
To help data center operators save energy and reduce operational expenses, the American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE) issued its Thermal Guidelines for Data Processing Environments. These guidelines suggest best practices among data center, HVAC, and facility operators to work together to reach an optimal IT environment. The guidelines include important information such as facility temperature and humidity measurement, equipment placement, and airflow patterns.
Technically advanced IT equipment functions efficiently in higher temperatures, which improves energy consumption and saves costs. However, many centers with new equipment are also running older machines as well. Therefore, carefully monitoring and making necessary cooling and heating adjustments will help to save money.
Humidity is temperamental in the data center environment. ASHRAE recommends a humidity level of 41.9°F dewpoint to 60% relative humidity, with an allowable range between 20% to 80%. When it’s too high, humidity can cause condensation, which can lead to corrosion and breakdown of components within the data center environment. Yet, when humidity is too low it can create electrostatic discharge across crucial components such as CPUs, RAM drives, hard drives, and motherboards.
Many external factors can change the internal humidity levels in your data center, especially if you are using an airside economizer system. When bringing cold air in from the outside to cool your data center, the air will need to go through a humidification or dehumidification process. A quality system will monitor your data center humidity levels from 0% to 100%. Humidity sensors placed properly throughout your server room will provide you with the information you need to monitor the varying humidity levels.
Temperature monitoring alone isn’t enough to protect your data center. Larger areas often require additional sensor information to determine potential problems with cooling efficiency. Proper airflow management ensures that cool air is going to the server intakes and the hot air emanating from servers is properly exhausted. Airflow management helps to control static electricity and dust build-up on computer fans that can cause equipment crashes. The sooner you learn of a ventilation failure, the faster you can act to prevent reduced server performance, equipment damage, and complete shutdowns. Airflow sensors notify you of any drop in airflow, which can lead you to the cause of the problem, like a clogged filter, failing blower motor, or unbalanced fans.
FIGURE 1. An example of an infrastructure monitoring system. (Image courtesy of Sensaphone.)
Often a water problem can be festering for some time underneath your raised floor, behind your walls, or above your ceiling panels before it becomes noticeable. Air conditioning units and cooling systems can leak from a failed condensate pump, broken supply lines, or clogged drains. Sprinkler systems can malfunction and pipes can burst. Properly placed leak detection rope sensors give you enough lead-time to prevent bigger problems.
Each server rack typically has an outlet strip powered by the power distribution unit (PDU), and each strip’s load is monitored by the PDU itself. The PDU has a built-in alarm panel that will trigger an output relay when a power issue is detected in any of these strips or the main PDU. Your monitoring system easily can be integrated into this alarm panel for instant alarm notifications. For redundant monitoring, an external power sensor can be plugged into each strip to monitor for voltage drops. This backup will give you the assurance that each strip has power. And if the PDU ever fails to trip its relay or send an alert, you can instantly get the information you need from the monitoring system.
FIGURE 2. An example of a cloud-based environmental monitor. Image courtesy of Sensaphone.
When power failure occurs, the uninterruptible power supplies (UPS) in your data center maintain the servers until the automatic transfer switch (ATS) calls for generator power. That’s why it’s important for you to maintain and routinely test UPS backup systems for optimum performance. A self-test is integrated in to each UPS. This self-test provides diagnostics on battery capacity, AC/DC load, and the charging system. It can also provide a power failure simulation. If any of these internal tests fail, an alarm panel will display the fault and activate a relay output. Equipping a monitoring system to the general alarm output on the UPS will produce an immediate alarm notification so that you can take corrective action.
Another key component to your backup power is the automatic transfer switch or ATS. The ATS transfers the load to your backup generator and can also advise the generator to start. Typically, there is a delay, and during this time the UPS units in place should maintain the power to your data center. Most transfer switches include auxiliary dry contacts specifically for alarming purposes. A monitoring system can seamlessly integrate into this alarm output.
Data center managers put into place many measures to preserve system security, but many are not protected against security breaches that can occur if an unauthorized person gains entry to the room after hours. Whether it is through an unapproved or forced entry or an accidental offense by someone on your staff, a breach of security could cause you to lose vital data, including proprietary information.
Data center monitoring systems can perform physical security functions. For example, you can add sensors to the entrance doors and the doors to the individual server cabinets. To avoid receiving constant alerts on busy days, you can configure the monitoring system to alert you only after hours and on weekends. You can also install passive infrared (PIR) sensor motion detectors inside and outside equipment rooms. They can interface with third-party IP cameras and your remote monitoring system. That way you can always get a first-hand look at the data center environment.
To complement your facility’s fire and suppression system, your remote monitoring system can detect the presence of smoke or fire within the data center, intermediate distribution frame, or PBX rooms and send a notification to the proper personnel. In addition, many fire alarm panels have auxiliary contact outputs that can be attached to your monitoring system for backup alarm notifications.
Base units, sensors, and peripherals
Each condition in your data center that you want to monitor requires its own input on the base unit of the monitoring system. You must match your needs with the number of inputs available. Taking an inventory of conditions prior to purchasing and installing your system is especially important if you need to monitor multiple sites with their own cooling and power systems. In addition to monitoring environmental conditions, some systems watch over IP devices, such as network links, routers, printers, and servers. When one of these components becomes unavailable, the monitor can tell you which IP device isn’t functioning.
Take an inventory of the sensors and accessories that come with your monitoring unit to determine what additional purchases you’ll have to make. Rechargeable battery back-up systems and power failure monitoring are often included. When selecting sensors, it’s best to choose ones that are designed by the manufacturer to work specifically with the monitoring system, and not universal components manufactured by a third party. If the components aren’t made by the system manufacturer, you’ll want to find out if they have been tested with the monitor you are choosing and if you need to work with another vendor to purchase the parts. For optimal functionality, it’s also important that accessories like switches, mount kits, antennae, power supply units, battery backups, and thermostats are made specifically for the unit.
Placement of sensors
Placement of sensors is another important element of a successful monitoring program. A temperature sensor located within an individual rack system will accurately monitor the temperature within that rack, but it won’t monitor the room temperature. If you are using multiple servers, computers, and racks, it is imperative that you place additional sensors around the room. Recommended temperature sensor placement includes next to the thermostat controlling the room temperature, two sensors on the front door of each rack, one sensor midway on the back of each rack, and one sensor for each air circulation plenum space.
Another example is the proper placement of a water sensor. Water on the floor can damage computer equipment beyond repair. Water tends to puddle at the lowest point of the floor and underneath pipe junctions and air conditioning units. If you place the sensor too high, it won’t trigger an alarm until after a serious flood has occurred. The entire perimeter under the raised floor should be fitted with leak detection rope, along with any pipes carrying fluids through your data center, especially under any valves or fittings within the pipe runs.
Wireless or hardwired sensors
Remote monitoring systems offer the option to have sensors hardwired directly to the base unit or sensors wirelessly connected. A hardwired monitoring system connects the sensors to the base device with wires. Generally, trenching long distances for wires is time consuming and costly. So alternatively, a wireless system uses built-in radio transmitters to communicate with the base unit. Some monitoring systems can accommodate a combination of hardwired and wireless sensors.
Communications to your site
Monitoring devices that use cellular communications must be registered on a wireless network (like Verizon or AT&T) before you can send or receive messages. Because cellular devices perform all communications over a wireless network, it is important that there be sufficient signal strength at your site. It is a good idea to check the signal quality in the area before purchasing a cellular product. If the cellular network has less than desirable coverage, it is possible to install an external antenna to help increase cellular signal.
When monitoring systems identify a change in status, they immediately send alerts to people on the contact list. If you don’t want all of your personnel to receive notifications at the same time, some devices can be programmed to send alerts in a tiered fashion. Make sure the system allows for flexible scheduling so that it doesn’t send alarms to off-duty personnel. It is important to consider the reach of the communications so that you’ll be notified regardless of your locations. Multiple communications methods like phone, email, and text provide extra assurance that you’ll get the alert. Also, note the number of people the system can reach and if the system automatically cycles through the contact list until someone responds.
Programming and real-time status checks
An important part of DCIM is receiving real-time information necessary to understand factors that affect efficient operation of your data center. Therefore, you want a system that will provide real-time status of all monitored conditions on demand. There are a few different ways to access your sensor readings. Options include calling to check status, viewing a web page, either on a local network or on the cloud, or accessing the information via an app on your mobile device. If you don’t select a cloud-based system, you will be limited to logging in through a local area network, which will allow you to make programming changes, access status conditions, and review data logs. With a cloud-based system, the devices supervise themselves. This means if the internet or cellular connection goes down, the device will send an alarm to alert the appropriate personnel.
FIGURE 4. A typical temperature sensor installed on a server room rack. (Image courtesy of Sensaphone.)
Data history is valuable in identifying patterns and trends in environmental conditions. Many monitoring systems automatically save information, recording tens of thousands of data points, dates, and times. Cloud-based logging provides an unlimited amount of records for users to view, graph, print, and export. Analyzing data samples may lend insight to larger issues and prevent problems before they arise. For example, if the temperature in a server room fluctuates slightly, you probably don’t have trouble lurking. However, if the data logs show steady temperature increases, it could be a sign that the cooling unit is having trouble keeping up with demand.
The monitoring system must be able to change or grow according to the needs of your facility. Be sure to factor in future computer room expansion, network growth, and equipment upgrades. Plan ahead by selecting a monitoring system with more inputs than needed, or choose one that can accommodate expansion. Planning for growth includes considering the number of devices and locations the system can monitor. Lower-cost, non-expandable monitoring systems could meet the needs of smaller facilities. But purchasing an expandable system could add value to the initial purchase because you wouldn’t have to replace the entire system if your facility expands in a few years.
Installation and support
Familiarize yourself with the installation requirements before you commit to buying the system. Make note of hardwiring, setup, programming, software setup, and integration requirements. Keep in mind the addition or replacement of equipment down the road. Ideally, technical support from the manufacturer will be available at the time of installation and throughout the course of ownership. Note that cloud-based monitoring systems don’t require users to install and integrate software, which can make the set-up process easier.
Return on Investment
When deciding how much you should pay for a remote monitoring system, tally up the entire cost, fully installed with additional peripherals and sensors and any labor fees for installation. Then consider the value of the equipment in your data center or server room. Next, factor in the cost of downtime should an environmental event damage your equipment and shutdown your operation. Add in the cost of repairing or replacing equipment. Include the value of your business reputation in the eyes of those affected by your system failure.
When it comes time to make the purchase, note that buying a remote monitoring system typically doesn’t involve a long-term vendor contract or even monthly fees. Look for a hassle-free purchase with no strings attached. By accommodating your needs now and in the future, your monitoring system can grow with your operation and likely outlive much of your data center equipment.