Posted on November 5, 2023 by Website Administrator
What happens when a data center overheats? Here’s a real-life case study: in October, major banks DBS and Citibank in Singapore experienced several hours of outage. Affected services included online banking, credit cards, digital payments, and DBS’s Automated Teller Machines in some locations.
In addition, Data Center Dynamics and Channel NewsAsia reported issues accessing Facebook, Whatsapp, and Instagram as well as Call of Duty servers across Southeast Asia, India, and even the UAE and Maldives.
All these disruptions stemmed from overheating of a few halls in the data center supporting these businesses. The operator, Equinix, stated that a contractor error resulted in a technical issue with the chilled water system.
While the Uptime Institute’s Annual Outage Analysis 2023 has found that outages have been falling in recent years, the outages that do occur are incurring higher costs, with over two-thirds of these costing over $100,000.
In this case, the cost may be on par or beyond that, with the amount of press and negative public sentiments generated for the affected businesses.
So how can data center managers prevent such incidents of under-cooling or other temperature-based issues from happening?
The answer: It’s time to deploy environmental sensors right. Because you can’t manage what you don’t measure.
Let’s start with temperature monitoring. Temperature sensors installed at a rack level could provide more accurate, real-time views of data center temperatures rather than computer room air conditioner (CRAC) readings alone. Additionally, temperature sensors can also be used to track temperature trends and help data center managers pre-empt potential hot spots.
It’s also important to note that installing temperature (and humidity) sensors should not stop at one per rack. Consider referring to the ASHRAE sensor placement guidelines to ensure your dashboard has readings from top, middle and bottom of the racks. While the investment needed for this may be a concern, the accuracy of temperature data can potentially save your team a significant amount of operating costs just days after installation!
Besides temperature sensors, airflow sensors can help managers monitor cooling airflow and hot air return on the racks, ascertain the cooling system is functioning correctly and the airflow is keeping the racks at the right temperature.
Have hot or cold aisle containment? Differential air pressure sensors can identify air pressure discrepancies indicating partition leaks in these aisles.
If the facility relies on an underfloor cooling system, underfloor air pressure sensors are essential for giving feedback to computer room air handlers (CRAHs), CRACs, or the building management system that alters fan speed to meet the underfloor pressure setpoint.
And how could we forget smoke, fire, and air quality sensors? These pick up on smoke, contaminants, and early signs of a fire, ensuring teams are alerted to take immediate action and minimize impact to other areas in the center.
Besides water-based chilling systems, moisture is usually unwelcome in data centers ? and there’s an abundance of humidity in tropical regions.
Humidity sensors help to monitor if the proper balance of humidity levels exist and help facillity managers avoid electrostatic discharges (when it’s too humid) and condensation problems (when it’s not humid enough).
Besides moisture in the air, having water sensors overhead and at ground level will ensure detection of any leaks, be it from overhead fixtures or pipes in water-cooled racks. Some may even alert mangers to the presence of a 50% glycol mix, which is a sure sign of a system leak.
Besides investing in the right sensors, consider if they are compatible with any existing monitoring and management systems in the data center.
If there is a compatibility gap in the centers’ setup, the facility manager will need to address it with intelligent sensor hubs or rack PDUs. These components will serve as the central connection point between the sensors, current infrastructure and a Data Center Infrastructure Management (DCIM) solution, ensuring the sensor data can be correctly archived, retrieved, and accessed by your teams.
The ideal device to connect sensors to the rest of your infrastructure should also support third-party sensors and can be installed quickly with the least amount of disruption to rack or area operations.
For new builds or expansions, the Raritan’s PX4 PDU, for instance, has an inbuilt controller with the industry leading Xerus technology platform that offers compatibility with many sensor types and integration to third party systems with the included Open API.
For existing environments the Raritan SmartRack Controller is a great retrofit option which also leverages the Xerus Technology Platform, supporting many different sensor types (including third party ones).
[See the impact an intelligent controller can have: Taiwan Telco Uses Raritan SRC to Enhance Monitoring of New 5G Cell Sites]
With the right cabled or wireless connection, a set of environmental sensors installed at strategic points, and the right monitoring tools, your teams will now have access to the environmental data , identify trends, and receive alerts if there is a change in the operating condition around the IT equipment.
Real-time data monitoring and measurements support the modern remote workforce, reducing the need for on-premise staff to perform redundant tasks that can be achieved by sensors. This also enables teams to chart patterns, generate reports easily with both recent and historical data, investigate any issues with more ease.
In addition, they can conduct predictive maintenance to minimize downtime, and reduce operational costs from emergency maintenance and repairs.
We hope this piece has given you a little more insight into how environment sensors can help data centers track environmental conditions and any changes thereof that could impact uptime, and potentially cause an outage like the one we recently experienced in Singapore.