Telecommunications carriers – operating in a business with low margins and plenty of competition -- must have the ability to swiftly collect and analyze customer data and take action based on it.
"The telecommunications business is a very data-intensive business," said Robert Geller, CIO with XO Communications. "You can't run a profitable business without being able to process and analyze vast amounts of data very quickly."
Like most telecoms, XO Communications, based in Herndon, Va., tracks each and every transaction made by its customers, mostly small and midsized businesses. Each call, text message and email generates data XO uses to set pricing rates, maximize billing revenue, and find the most efficient way to route traffic over its network.
Until 2008, however, XO's IT infrastructure was nearly overwhelmed by the vast amounts of data the company collected on its more than 90,000 customers. Processing and analyzing data took days or even weeks, meaning that XO was often late reacting to changing market conditions.
"We were struggling," Geller said.
Determining the optimal routing path and prices for long-distance calls from one city, for example, would take XO's servers and databases nearly a full day to complete. With XO covering more than 75 metropolitan areas in the U.S., it took months to set long-distance routes and prices for the entire country. By that time, demand and network conditions had often changed, making the recommendations quickly out-of-date.
By 2008, Geller decided a change was needed if XO were to remain competitive with its larger, more established telecom rivals like Verizon and AT&T.
He initiated a proof of concept, inviting a number of data warehouse vendors, including Teradata and Greenplum, to compete for XO's business. Both did well in the proof of concept, Geller said, proving adept at -- among other things -- supporting large numbers of users.
But Geller was more concerned with finding the warehouse that could best handle large data volumes, not multiple users, as only a handful of XO employees would be using the system once deployed. For that, Geller was most impressed by a data warehouse appliance from Netezza.
In addition to its ease of use, the Netezza 10,000 Series appliance's ability to handle large data volumes and dramatically shorten the time it took to complete a number of XO's analytic processes ultimately won the Marlborough, Mass.-based vendor the business, Geller said.
XO's situation is hardly unique. The amount of data generated by mobile devices and smartphones is expected to explode over the next decade. U.K.-based analyst Informa predicts that mobile data traffic alone will see a 25-fold increase by 2012.
As a result, telecom carriers will increasingly turn to sophisticated, multi-terabyte – even multi-petabyte -- data warehouses to help them make important business decisions, according to James Kobielus, an analyst with Cambridge, Mass.-based Forrester Research.
In fact, many telecoms have already made the move, Kobielus said. Most of the largest current data warehouse deployments in the world are at telecommunication carriers, he said, and they're only going to get larger and more sophisticated.
"Today's data warehouse environments will look painfully primitive by 2020," Kobielus said.
The advent of massive parallel processing (MPP) and the adoption of cloud-based deployments, solid-state drives and in-database analytics will allow data warehouses to continue to scale to larger and larger data volumes, he said. And telecoms will be among the companies that will benefit most.
But XO is enjoying the benefits of its data warehouse appliance in the here and now. Since deploying the appliance at the beginning of 2009, the company has reduced the amount of time it takes to produce a nationwide, long distance pricing plan from weeks to just a matter of days, Geller said.
That's because the appliance's MPP architecture allows it to process multiple data sets simultaneously, rather than one at a time, according to John Gillespie, vice president and general manager of global telecommunications at Netezza.
In fact, the appliance, which cost XO $1.4 million, paid for itself in less than two months, Geller said. It even helped one XO division go from a loss of $2 million one month to a $2 million profit the next. And the company intends to expand its use to financials sometime this year. XO is currently using only about half of the appliance's 50 terabyte capacity, according to Geller.
"It's a game changer," he said. "These days, it's so competitive and the margins so slim, if you can't use the data effectively, you're just not going to make money."