Datacenter Outages: Comparison of Federal vs. Private Sector

In the wake of the 2013 healthcare.gov failure, the private sector came to the rescue of the federal government. If anything, the incident brought to light the massive chaos that can be caused by sub-standard application development of a site intended to run uninterrupted in massive data centers and expected to support millions of users. It would be interesting to see how the private sector compares to the federal government in terms of data center performance downtime and cost.

Two 2013 Pomenon Institute studies1,2 , sponsored by Emerson Network Power, paint a very detailed picture of the data center downtime cost for the private sector. The counterpart survey of the federal sector by MeriTalk3, sponsored by Symantec, came out in August 2014. It examines the cost of government data center downtime on federal employee productivity. A point-by-point comparison of the two studies is difficult due to the different methodologies and questions in the two surveys. Here we compare private and federal data center performance on 4 metrics that are presented in both studies and give an idea of the differences and similarities of the two sectors.

(NOTE: The private sector surveys are based on incidents over a two-year period, while the federal one on incidents over one month. We accept that that the one-month figures are representative of a longer period of time.)

1. Average incident duration: 86 minutes for the private sector, 91 minutes for the federal. The two figures are very close to each other.
2. Cost per minute of downtime: The private sector loses $8,026 and the federal government $1,956 per minute of data center downtime. These first two points bring the average cost per incident to $690,204 for the private sector and to $178,000 for the federal government. This is expected since the private sector data centers contribute to productivity as well as revenue, while for the federal government data centers contribute mostly to employee productivity. It has been estimated that federal employees save 800 hours or $32.5B annually by using applications and services that run on government data centers.
3. Root cause of unplanned outages: There is a different set of reported root causes of data-center outages between the two sectors*, but the human error and weather are two factors that appear in both studies. We see that human error is a much bigger problem in the private sector (22% vs. 6%), while weather related incidents are practically equal in both cases (12% vs. 13%). (*We don’t know if the participants were given a list of root causes or they freely reported their answers)

4. Equipment investments: 49% of the private sector and 56% of the federal government respondents listed equipment investment as the top way towards truly reliable and agile data centers. We see that the numbers are almost in agreement in this question. It is not surprising that to decrease unplanned outages and consequently contain losses, it boils down to investing in infrastructure and technology.

All industries are striving for higher and higher network and data center uptime, and at the same time, they hurt more and more from unplanned outages. The government and the private sector have very similar goals driven by different needs (productivity vs. revenue). We tend to think the government as a late adopter of new technologies and trends, but given the recent revolutions in networking and data-center technology, nobody can afford to stay behind the curve. There is far more data and surveys that track trends over a long period of time for the private sector than in the government. It will be interesting to see ongoing publication of these surveys in the coming years in order to see how the private and the federal sectors evolve.

1. Ponemon Institute, “2013 Cost of Data Center Downtime,” September 2013.
2. Ponemon Institute, “2013 Cost of Data Center Outages,” December 2013.
3. MeriTalk ,”The Drive to Thrive: Ensuring the Agile Data Center,” August 2014.