NK

NK Transforms Big Data into Actionable Insights that Drive Business Success

Highlights

NK (NK.pl) is a Polish social networking service that brings students and alumni together online. Created in 2006 by 24-year-old Maciej Popowicz, a computing student in Poland’s high-tech hub of Wroclaw, NK is an abbreviation of "Nasza Klasa" which literally means "our class". The site has been compared to Britain’s Friends Reunited, France’s Copains d’avant, and the US’s Classmates, and Facebook. Yet NK.pl is more popular than Facebook in Poland, boasting more than 13 million active users compared to Facebook’s 6.5 million (in Poland). According to Google’s annual Zeitgeist list of commonly used Web search terms, NK was the seventh-fastest rising term in the world in 2008. The success of NK.pl has been attributed to its high-quality social networking features, online gaming offerings, and targeted advertising capabilities.

The social networking site with the widest coverage and penetration in Poland, NK.pl strives to provide consistently excellent experiences to its large, diverse base of users—in addition to helping advertisers hone ad effectiveness for better ROI. To stay on top of constantly evolving members’ tastes, interests, and requirements, NK product managers collect and analyze click-stream data from a variety of sources including most widely used features, most popular games played, and ad click-through rates. Insights gained from this analysis drive feature improvements, advertising revenue, and new product offerings.

Challenge

In this digital age, trends and tastes seem to change at the speed of light. In the past, NK.pl had trouble keeping pace with members’ changing preferences and behaviors. That’s because NK’s traditional transaction database was simply not powerful enough to process the 250 terabytes of data generated by 13 million users in a timely manner. In fact, when product managers attempted to run queries against the data—which was spread across 1,500 servers—they faced query response times of three to four weeks.

Such slow response times were unacceptable to NK.pl because the data was outdated and meaningless by the time it was received. To optimize product features, marketing efforts, and the experience of its social network members, NK.pl needed a system that could perform both simple and rich queries quickly for intra-day analysis and response in an extremely cost-effective manner.

Solution

In an effort to speed query response times, NK.pl built a data warehouse that would open new windows into data stored in the transaction database. Split into two main parts, the main "fast" analytics portion now sits on a single server and is powered by the VectorWise analytic database. The other main part contains large data storage and analytics powered by Hadoop solutions. Hadoop technology stores click-stream data, such as POST and GET requests as well as AdServer logs. Before integrating Vectorwise into the data analysis infrastructure, the data warehouse could not deliver the query response times that the organization required. Now, employees use Vectorwise to achieve "speed-ofthought- analysis" for most important queries which are made on transactional database database extracts in addition to extracts from large data storage. According to Edward Mezyk, Senior Project Coordinator in NK Research and Data-warehouse Division, "We looked to solutions from other vendors with analytic databases, but selected Vectorwise for its superior performance and cost-effective model."

Results

Today, NK collects an additional one terabyte of information per day from its servers, including user attributes, messages, comments, and photos. With Vectorwise, product managers now run 50-90 huge queries daily and receive sub-second response times. This helps them gain fast, actionable insights to make more fully informed strategic business decisions. "Our business runs on knowledge," says Mezyk, "Vectorwise delivers timely knowledge that will help us drive revenue today and into the future."