When Vodafone CEO Vittorio Colao said on November 20, 2013, his company was considering keeping its annual capex above traditional levels to fund the rollout of next-generation fiber and 4G networks, Vodafone lost about $3.7 billion in market capitalization on that day. While higher-speed networks delight customers, they do not go down well with shareholders obsessed with cash flow and dividend yield.
Fortunately, Hadoop can stretch necessary network investments by a modest yet meaningful amount. If a large Hadoop project costing $1 million dollars could delay or reduce by just a few percent annual network capex spend, which for the large operators tops $1 billion per year, what’s not to like?
Let’s look at four examples where leading carriers are deploying Hadoop today to optimize their network investments.
Network capacity planning benefits from better and more timely operational intelligence. The consumption of services and resulting bandwidth in a particular neighborhood may be out of sync with a telco’s plans to build new towers or transmission lines in that same neighborhood. This leads to a mismatch between expensive infrastructure investments and the actual revenue from those investments. By analyzing Call Detail Records (CDRs) and network loads, telcos can plan infrastructure expansion with greater precision. “One European carrier used Hadoop to optimize the rollout of 4G coverage in time and space to match the likely pick-up in service revenue, based on detailed cell tower traffic data of the last few years,” Peter Weichsel, CEO of P3 Digital (a branch of P3 Group that regularly evaluates big data technologies), said at Big Data Monetisation in Telecoms in November 2013. “With their prior, less informed approach, they would have had to spend 10 to 20% more capex for the same outcome.”
A US carrier is using Hadoop to improve cell phone service. This operator has had a sophisticated, end-to-end global network management platform that enables it to understand and fix instances of poor cellular service such as dropped calls or poor audio quality. However, because of the tremendous volume of data, coming in at an average of 10 million messages per second, and because of the high cost of existing analytic solutions, each analysis was limited to a 24-hour time window and only one-fiftieth the surface area of the United States.
This creates a “groundhog day” scenario for customers. The same customer issue may generate multiple support calls, but the operator’s team cannot see relationships between multiple variables across time. Is the problem with the customer’s device? Is it their neighborhood or proximity to a tower? Is it because of how they use their phone? Or is it all of the above? Rob Bearden, CEO of Hortonworks, told me, “With Hadoop, this carrier can afford to retain all messages for a rolling six-month period. With more history, they are able to explore root causes that they have never been able to identify by reviewing just one day’s data.” Hortonworks’ distribution of Hadoop is deployed at seven US carriers.
Hadoop is also being used for targeted network maintenance and upgrades by cable companies. One large US cable MSO was unsure how cable network congestion affects churn, and where exactly network upgrades produce the most incremental revenue. The answer lies in analyzing node utilization against customer experience indicators to see if congestion correlates to the end user’s experience.
That includes looking at the increase of dropped packets, any increase of calls into the call center for customers associated to these nodes, and any increase of requests to drop service. “Only Hadoop was up to the daunting challenge to correlate and make sense of 4 million subscriber records, 12 million work orders, 9 million contact center calls, 42 million IP detail records and 20 million Tivoli NetCool Performance Manager records. The result: only a small number of nodes were responsible for the majority of the negative customer experience,” says Ben Sharma, CEO of Zaloni, a company that provides network analytics and data management solutions based on Hadoop.
Hadoop also helps with real-time bandwidth allocation. Certain mobile apps and user activities can hog bandwidth and erode service quality for all other customers, perhaps because they contain malware from non-trusted app stores. Network operators need to respond to such bandwidth spikes quickly to reallocate virtualized resources and maintain service level agreements (SLAs). With a combination of real-time deep packet inspection and text mining the transcripts from contact center support calls, operators can steer traffic and optimize network quality of service (QoS) in real time in an attempt to maintain the best service quality for the largest number of customers.
These communications network examples show what enterprises across industries are discovering: Hadoop brings both superior economics compared to legacy analytics, data warehousing and storage alternatives as well as exciting new capabilities. These capabilities provide deeper and more actionable insights to drive revenue up and costs down.
Juergen Urbanski is a board member for Big Data and Analytics at the German IT industry association BITKOM. Gigaom’s Structure:Data event, held March 19-20 in New York City, will delve into more strategies for big data.