Data: Go Big or Go Home

The amount of data in the cable architecture that potentially could be tapped is ... well, big. Yet, until recently, utilizing it would have been difficult because of storage costs and processing speeds. As the former has dropped and the latter risen, however, "big data" metrics could feasibly be leveraged to help operators with anything from distribution plant issues to increasing ad revenue."We are at a point where companies are storing data even if they don't know how to use it," said Kip Compton, VP and CTO of Cisco's (NASDAQ:CSCO) video and collaboration group. "They will figure it out and use it in the future." Compton co-authored a paper that was presented at the Cable Show titled, "Big Data: Capitalizing on Untapped Knowledge."The main differences between "traditional data" and big data are size and structure. Consider this: While databases generally hold between several gigabytes and several terabytes, big data requires storage capacity of more than a petabyte. Also, the concept stemmed from Google's (NASDAQ:GOOG) work on searching the Web. This means that manipulating big data involves multiple attributes and formats, as well as correlating across data sets to look for patterns, Compton said.From the cable perspective, data sources could include set-tops and other clients, the cable plant itself, the content delivery network (CDN), and the customer call center, for example. The big data that is mined could be used in the field (to predict, diagnose and correct problems perhaps before the customer is affected), to determine viewership (which channels should be switched vs. broadcast), to target advertising (content affinity can be used for cross-selling), and capital investments (what needs to be upgraded and when), the report says.A big data architecture will be different depending on what information interests a cable operator, how often they want to receive it, whether the collection point is a distributed or centralized database, and who needs to see the data, said Art Howarth, director of engineering, video and collaboration group, Cisco, who presented the paper at the Cable Show."It is (also) device dependent on how you extract data from it ... the call center is different from the CDN," Howarth said. "Operators should start with a specific problem statement and expand off of that."While a lot of operators will start piecemeal and build as they go, common analytical tools are important. "A lot of insight can come from the correlation of data between what the set-top and the cable plant say," Compton said. "If you are storing and analyzing data in completely ways, it can be difficult (to get the complete benefit)."The basics are the need for the instrument to generate data. The set-top, for example, requires code so that data will be captured and sent back to the operator. There has to be a repository or storage architecture for keeping the data and an analytics platform that uses the information to generate insight."The real gem of all this is the visualization tool," Howarth said. "Big data is a lot of responses. It could be reams of data. The ability to visualize is (important). The eye can catch trends better visually than off of a layered spreadsheet."Monta Monaco Hernon is a free-lance writer. She can be reached at [email protected].