We live in a world preoccupied with numbers. Whether it’s following a favorite baseball player’s batting average, temperatures for this week or how many miles till the next oil change, numbers run our lives.
The same holds true, of course, in data centers. The most popular number there seems to be 99.999—otherwise known as “the five nines.” In a recent article in Data Center Dynamics, Cliff Federspiel, founder, president and CTO at Vigilent, a pioneer in dynamic cooling management in mission-critical environments, says this figure has become more of a marketing statement than an accurate assessment of overall data center reliability.
The concept of five-nines’ uptime—a decade or more in the making—typically referred to internal components, such as network switches, routers and power components. Nothing has changed over the years when it comes to organizations struggling to achieve the gold standard of uptime.
According to the results of Uptime Institute’s 2019 Data Center Industry Survey, more than one-third of respondents experienced an IT service outage or severe service degradation during 2018. What’s worse is that one in five of the outages was rated as “serious or severe.”
Most troubling is that the highest percentage of outages—33%—were blamed on on-premises data center power failures. This hits particularly hard for us at VPS because we know that the implementation of Software Defined Power (SDP) can reduce this number significantly. What’s required is a new way of thinking and looking at power—from a software point of view.
Adopt a Software Strategy
To get started, we must move away from the decades-old reliance on trying to achieve 99.999 uptime by continuing to overprovision power. Instead of being constrained by power—60% of power capacity typically is not utilized—we must take a different approach to defining and achieving data center reliability.
Instead of putting more resources into adding reliability and redundancy into hardware infrastructure, adopt a software strategy. Through SDP’s federation and fault avoidance using predictive analytics, data centers can increase rack power capacity and utilization of available power while fully automating operations for better efficiencies.
Some diehard “five-nines” seekers might feel that building fault tolerance into every inch of infrastructure is far preferred over deploying software. Especially in the power management space, the concept of adding a layer of software intelligence challenges traditional thinking. But the reality is that simply throwing hardware at any problem is not the best possible fix.
Clearly, the numbers speak for themselves—the approach taken for the past decade hasn’t brought data centers any closer to achieving maximum levels of reliability and resiliency. It’s time for a different way of thinking.
It’s also time to debunk the misconception that adding SDP requires a lot of change to existing data centers. This simply isn’t true. In fact, SDP capabilities can be phased in over time, starting with the opportunity to gain visibility across the facility without making any hardware changes.
The opportunity to gain power awareness is the first step to understanding what’s really happening while facilitating faster recoveries. With SDP, more granular visibility basically comes for free—and you don’t need to redesign your data center to get that advantage. There’s much more you can do beyond that in terms of adapting your business model to take advantage of this newfound awareness and flexible control of power resources.
Of course, the biggest payoffs of this visibility is increased reliability, lower failure rates and better business outcomes. Simply put, SDP helps solve the problem instead of focusing on the symptoms.
The Reality of Reliability
Data centers would all be more effective if their focus shifted to achieving the highest possible application availability. The challenge today is we need a new number. That is something we are trying to do with Software Defined Power. The entire data center industry needs to create a new concept, a new number that organizations can use to anchor their data center and power infrastructure designs and strategies.