Taking data center monitoring to the edge

Taking data center monitoring to the edge

Data center monitoring

The future is now, and nowhere is the proof more evident than in the shift toward edge computing.

Data centers and server rooms deployed at the edge of networks bring cached versions of digital content that live on origin servers closer to end users in specific regions. Why is this necessary? The short answer is that the speed of light, while fast, can't deliver sub 10-second latency without the help of locally stationed IT infrastructure, according to Data Center Knowledge contributor Cole Crawford.

This is crucial because most cutting-edge technology trends demand extraordinarily low latencies for very high volumes of data. These include, but aren't limited to:

  • The Internet of Things (e.g., smart homes, healthcare sensors and other "intelligent" endpoint deployments)
  • Augmented and virtual reality
  • Autonomous industrial automation systems and vehicles
  • Machine-learning
  • The increasingly real-time rate at which we expect data to be processed

Edge infrastructure represents the beginning of a new digital age – one where "real-time" describes data that is processed up to the millisecond. The more dependent on edge data centers we, as humans, become, the higher the stakes should an environmental adversity induce an outage on an edge server.

Consequently, there has never been a more important use case for data center monitoring than at the edge of modern networks. Facility management needs to know that something could go wrong the second that information becomes available.

Edge infrastructure now helps process data at hyper-quick speeds.Edge infrastructure now helps process data at hyper-quick speeds.

Life on the edge demands vigilance

Facility managers who aren't careful risk heading straight off a different kind of edge: the edge of their data center's environmental limitations.

The American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE) recommends maintaining ambient temperatures of 64.4 to 80.4 degrees Fahrenheit. Some CPUs can operate safely at temperatures of up to 85 degrees F, but according to ConsumerPC, most nodes that operate any higher than 86 degrees F for a few minutes risk getting fried. This would be nothing short of a disaster in an edge data center, especially one that regularly sees high traffic and volatile utilization spikes. 

This raises two points:

  1. If your ambient temperature has exceeded 85 degrees F, you're too late; you need to catch temperature fluctuations before they hit a point of no return. 
  2. If you want to let your equipment run warm (<80 degrees F), bear in mind that a single CRAC failure or airflow deficiency puts you much nearer to that precarious 85 degree F cliff. This makes real-time temperature monitoring and targeting even more important. 

To preempt a meltdown, adhere to ASHRAE's recommendation to monitor temperature at six points in each rack – three sensors for air intake and three for air outtake.

Also, be sure to monitor humidity and dew point. Warmer air can hold more moisture, which means the inverse is true for cooler air. Failure to account for relative humidity could result in condensation on equipment that, in turn, may cause long-term corrosion or, worse, near-term fried circuitry. 

Most importantly, leverage automatic alerting via SNMP, SMS email or other means to make sure that you actually know when an environmental threshold is in jeopardy. With remote environmental monitoring, all you need is a device and an internet connection to receive real-time alerts.  

Wireless monitoring is better

Space is a commodity in any data center, and the same goes for edge facilities. Ideally, you want to deploy environmental sensors that wirelessly connect to the local network through a dedicated gateway device. This saves space, but also expedites installation and makes it easier to scale out your environmental monitoring infrastructure. 

For context, a best-in-class wireless data center monitoring system will support about 400 sensors on one gateway and have a maximum transmission range of approximately 1,000 feet. This ensures that even the sensors at the far edges of your facility will continue to transmit data in real time.

Clearly, there's more than one way to take data center monitoring to the edge.