Data centre outages increasingly caused by DDoS
21 January 2016 | 0
Think housing your servers in a data centre rather than squeezing them under your desk is a bulletproof solution?
Well, they might be safer in a data centre, but believe it or not, some of the same pitfalls that can create trouble in the office can affect those secure data centres too. Namely UPS failure, human error, and cybercrime.
UPS system failure is still the principal cause of “unplanned data centre outages,” according to a new report.
A quarter of all such events were related to UPS systems and batteries, according to Emerson Network Power in association with Ponemon Institute. The two organisations have been studying the cost of unplanned data centre outages.
But cybercrime-caused outages, specifically Distributed Denial of Service (DDoS) attacks, constituted a whopping 22% of the unplanned disruptions last year. That’s up from just 2% in 2010 and 18% in 2013, the last times the two organisations performed the survey.
The survey collected responses from 63 data centre operations who had observed an outage in the prior about year about what exactly happened. The report was published this month.
“Generators appear to have become more reliable. Those systems contributed to 10% of the failures in 2010, whereas today they only make up 6%”
Accidental causes or human error were the third biggest cause of unplanned outages, according to the report. Those mishaps caused 22% of the failures. That is the same percentage as in 2013, but lower than in 2010, when 24% of outages were accidental or human-caused.
Interestingly, many other causes of outages are lower now than they were in 2010 and 2013. They’ve been usurped by cybercrime’s huge gain.
UPS failure is down slightly on 2010, when it accounted for 29% of the outages. And the aforementioned human error is down a bit. And utility failure, such as water, heat, and Computer Room Air Conditioning, which today makes up just 11% of the outages, was at 15% in 2010.
Likewise, generators appear to have become more reliable. Those systems contributed to 10% of the failures in 2010, whereas today they only make up 6%.
The researchers do not provide numbers relating to changing data centre design over the period. Fewer generators in use — replaced by solar and alternative energy — could conceivably have caused that statistical decline. The report does not specify.
Overall, most unplanned outage causes — including those caused by weather, which accounted for 10% of outages this year, compared to 12% in 2010 and 2013 — have declined in favour of cybercrime.
Even IT failure, a measly 4% of failures today, dropped from 5% in 2010.
$9,000 per minute
And the cost? The report was released to expound on the cost of the outages, rather than to apportion blame.
Well, the “average total cost per minute of an unplanned outage increased from $5,617 (€5,157) in 2010 to $7,908 (€7,260) in 2013 to $8,851 (€8,126) now,” according to the report.
Downtime at data centres now costs an average of $740,357 (€679,705). That’s a 38% increase on 2010, the study calculates.
And maximum costs are even higher.
“Maximum downtime costs are rising faster than average, increasing 81% since 2010 to a current high of $2,409,991,” (€2,212,000) the report says.
Patrick Nelson, IDG News Service