The Numbers Don't Lie

To cut costs, Statistics Canada adds a partner and consolidates storage

	Stats On
Statistics Canada Founded: 1918 as the Dominion Bureau of Statistics Employees: 5,700 IT staff: 1,200+

It was the American author Mark Twain who, in the late 1800s, popularized the statement, "There are three kinds of lies: Lies, damned lies, and statistics." More than a century later, it took just a few Canadian statisticians and a bit of technology to prove him wrong.

In 2002, the Canadian government issued a nationwide directive for its agencies to reduce operating budgets. Among the agencies obliged to follow this directive was Statistics Canada ("StatsCan" for short), the organization that conducts a national census every five years. It also collects and analyzes statistics on everything from population growth to small business performance.

In response to the mandated budget cuts, Ivan Fellegi, StatsCan's chief statistician (the equivalent of the agency's CEO), challenged the agency to streamline operations. This challenge posed a problem for the agency's informatics technology services department (ITSD), which was experiencing an increased need for storage.

In 2001, Canada offered some citizens the choice of responding to the census over the Internet, rather than on paper.

"We got 7 terabytes of data from that," says Guy Charron, assistant director of StatsCan's infrastructure services. In the 2006 census, all 19 million Canadians old enough to respond had the option of providing their data online, and about 20 percent did so. Australia and New Zealand also offered the public the chance to respond online to a 2006 census, but neither country had a higher return rate than Canada, according to the U.S. Census Bureau.

The new census method means huge storage needs: Charron's team estimates it will need 50 terabytes of storage from that 20 percent census response alone. The storage will need to be long-term, as the Canadian Parliament mandates census records be kept for 92 years.

StatsCan needed to increase storage dramatically while simultaneously cutting costs. Charron believed it was possible-if the agency could consolidate its data backup and storage into one stack overseen by ITSD.

In 2003, StatsCan had about 100 terabytes of data, most stored and managed locally by eight different operating divisions on more than 750 servers. That year, ITSD conducted a total cost of ownership survey, collecting information on data storage agency-wide. The survey showed that 49 percent of disk space was unutilized. Because backup and storage use was rapidly growing, it also suggested that these would be the principal drivers of future IT costs.

Though ITSD had been offering centralized backup and storage as an option to all of StatsCan since 1999, only 31 percent of stored data had been migrated to ITSD servers.

When Fellegi challenged ITSD to streamline, Charron and his team responded with a proposal to mandate that ITSD handle all backup, storage, and file and print services for the entire agency. "The initial business case showed savings of close to CAD 3 million [about US$2.7 million] over five years," Charron says. He knew that figure would be even greater once StatsCan was able to realize IT labor savings due to consolidation. "That's when we started making the case that there were economies of scale."