37% of Data Center Outages Most Likely Caused by On-Site Power Issues
In a sector that is growing and adapting to rapid change over the last year on multiple levels whether outages, resiliency, staging, or deployment of innovation, the Uptime institute’s latest Annual Outage Analysis 2021 & Global Data Center Survey highlighted key challenges for powering a data center with ongoing trends that identify the need to review of how mission-critical the electrical infrastructure is for the industry.
- Over a third of major outages are most likely to be caused by power failures and impact everything on-site and can cause knock-on effects to systems, software & networks
- Bigger outages are becoming increasingly more damaging and expensive
- Outages continue to occur with disturbing frequency
Uptime Institute survey findings showed that the frequency and impact of outages continue to occur but are becoming more damaging and expensive. This trend in their findings over the last three years hasn’t diminished but has become more apparent. Alarm bells should be ringing. Power problems are still the biggest cause of major outages.
In Uptime’s survey, on-site power issues caused 37% of noteworthy outages. The impact of a power outage is wide and deep, and the knock-on effects can be long-lasting — even if the initial failure is quickly fixed.
Avoiding downtime remains a top technical and management challenge for all owners and operators. So, while data center infrastructure management systems and the introduction of artificial intelligence-driven services have proliferated around systems, software, and network, more needs to be done to protect power infrastructure.
The case for continuous monitoring and more automation is increasingly compelling when it comes to the condition of electrical assets.
Over 75% of the Operators who responded to the Uptime Survey agreed that with more attention and investment, outages would almost certainly fall off significantly, yet for maintaining electrical assets, the move away from periodic inspection (typically annual), has not accelerated as quickly despite engineering research that indicates 24x7 Condition Monitoring will predict 70% more failures in advance than traditional inspection.
Perhaps it’s a lack of practical and shareable insights on how to specify the right sensor technology, why it’s of critical importance to enable reliable data collection and convert it into actionable data to effectively automate systems and maximize obtainable cost benefits?
It is a given that in a changing world with demand to live and work remotely increasing that compute-intensive workloads for data centers will grow exponentially — many of these will significantly push up power use, rack density, and heat.
This puts pressure on owners and operators and raises questions on DCIM and BCP best practices to innovate the way more complex electrical infrastructure is maintained and overcome the challenges of energy use as it continues to rise.
Our upcoming series will cover these topics and more as we share some of our tips to improve the process and engineering for data centers.