Apache: Big Data 2015

As dominant social network service provider, Daumkakao confronted several technical challenges to store and traverse large graph data.
First, our social network has 10 billion edges and 200 million vertices and users create 1 billion new edges everyday by interacting with our services so our system needed to be distributed and scalable.
Second, our system needed to provide low-latency and high concurrency to meet our quality of service.
Third, for viral effect, user`s activities should be delivered to right place at right time in real time so simple result cache can`t be used.
Lastly, there are about a hundred of services that daumkakao is operating and our system needed to provide common way to store and traverse data for synergy between services.
S2Graph successfully solved these technical challenges, so we'd like to introduce the methodology and architecture we used.