The actual cost of datacenter downtime

The actual cost of datacenter downtime

Summary: The cost of failure is high, and identifying the potential problems is the first step towards preventing them

TOPICS: Data Centers

It should come as no surprise to anyone that as cloud and network based services become more prevalent the impact on business when these services fail has become more pronounced. According to a survey from Ponemon Institute, sponsored by Emerson Network Power, the actual cost, in terms of direct, indirect, and lost opportunity, has increased more than 40 percent in the last 3 years.

More than 90 percent of the 450 datacenters professionals surveyed reported that they had experienced an unplanned outage in the two years prior to the study. On average, they reported two complete datacenter outages in the study time period, with an average downtime of almost two hours. Based on the Ponemon Institute calculations, these two hour outages cost an average of just over $900,000 or more than $7900 per minute. Partial outages, which were defined as the outage of one or more racks within the datacenter, had an average recovery time of less than one hour with an associated cost of about $350,000.

More than 80 percent of the respondents were able to identify a specific root cause of their system failure, reported as follows:

  • UPS battery failure (55 percent)
  • Accidental / human error (48 percent)
  • UPS capacity exceeded (46 percent)
  • Cyber attack (34 percent)
  • IT equipment failure (33 percent)
  • Water incursion (32 percent)
  • Weather related (30 percent)
  • Heat related/CRAC failure (29 percent)
  • UPS equipment failure (27 percent)
  • PDU/circuit breaker failure (26 percent)

The most positive aspect of the report is that in the three years between studies (the first was done in 2010), datacenters, on the whole, have become more reliable, with both the number of incidents and their duration going down.  But with the nature of the cloud-based business model, both public and private, the reliability of your datacenter servicers and the supporting infrastructure will become exponentially more import to the business bottom line.

Topic: Data Centers

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.


Log in or register to join the discussion
  • You can protect yourself.

    Minimize your usage of these central datacenters by allowing as much productive work as possible to be performed with offline copies of data. Of course, this is not always possible (which is why airlines, with a perishable commodity that had to be sold, but not oversold, from many different "store" locations, hires computer companies to develop 24/7 remote access to a centralized database, what we call a "cloud" today, back in the 1950s). But if, say, 80 percent of the centralized data needed by a specific user on a specific device can be kept on that device and periodically synced with the centralized copy, and if changes to that data by other users on other devices are relatively rare, with corrections not critical, then it makes sense to "fail softly" by using the currently locally available file for doing business or personal chores when the cloud is not available, and reserving the right to make corrections after the cloud can be accessed.

    In the 1950s and through the end of the century, desktop terminals had little or no ability to make local copies of files at a remote location and work with them, because they could only add commands to screens, or fill out forms on screens, and send them to the "big iron" for processing. But really, now that the technology EXISTS to run a full featured file storage and computing system (a mini datacenter) in a desktop, laptop or even tablet, why go back to the past by putting ALL the computing capability (such as the ability to update a spreadsheet or print pages from a document) in the "cloud" and turning your individual device into, effectively, a 20th century "dumb" terminal? This makes it impossible to do ANY work when the "cloud" server is down.
  • Correction to above:

    ... the airlines HIRED computer companies ... in the 1950s.