“Those who cannot remember the past are condemned to repeat it.” – George Santayana
From a distributed IT and data center monitoring perspective, what if cloud-based data center infrastructure management (DCIM) software could help you determine whether current events are critical or unimportant based on historical data?
Knowing the difference between a non-event, an emergency, and an outright crisis is vital to continuous and efficient operations. IT and Data Center Professionals must be able to differentiate between the thousands of alarms that can be activated at any time in a normal working day.
Different level events must drive different actions
Operators can face many challenging scenarios and must make quick decisions based on real-time data. They must learn to classify the impact of alarms to determine a proper response.
Software with predictive analytics could take routine events – like an alarm firestorm – and prioritize which events are true emergencies.
Other events are unique – so extreme they are impossible to predict based on historical information.
Then there are those events that are difficult, but not impossible to predict.
In the European heat wave of 2018, one data center operator reported getting 20,000-30,000 alarms a day because the outside air temperature was regularly rising above 35°C (95°F). Operators quickly recognized a pattern. An uptick in temperature routinely happened before the cooling system kicked in and reduced temperature. This repeating cycle caused temperatures within the data center to regularly rise above, and then fall below, target thresholds as often as 30 times a day. Monitoring systems responded by generating thousands of alarms on the rack PDUs as the thresholds were crossed. Analysis helped operators understand which alarms needed attention and which could be ignored.
A software system that can recognize the characteristics of an event based on historical data, and grade the effects and responses to previous alarms, can inform the personnel about how critical or not such an event is likely to be.
Stop reacting, start predicting
New cloud-based data center infrastructure management (DCIM) tools are constantly evolving to improve operations and mitigate potential issues. By capturing endless, anonymized data points, cloud-based DCIM software is designed to help operators move from being reactive to being predictive.
The software system learns from operator’s reactions to different events and becomes “smarter” by establishing alarm priorities based on previously seen conditions. By automatically building ratings of criticality each time an alarm occurs, the software can guide operators’ reactions with appropriate warnings. If a crisis does occur, the system stands ready to highlight what is important and push to the background anything that is not. Through this support, operators can more quickly perceive what constitutes the real cause of a problem and respond appropriately.
The software system can become personalized to a particular IT environment and those who operate it in much the same way that a search engine, online retailer, or digital content provider learns your preferences and serves items that relate directly to your primary interest.
Discover today’s cutting-edge cloud-based DCIM tools
In a distributed IT or data center environment, we are always learning more about real problems. Improving today’s alarm systems will increase productivity and help deal with major hazards. As systems scale and more data is captured, predictive analytics become feasible. This has the potential to raise the bar on what is possible through state-of-the-art, cloud-based DCIM software.