Big data applications require us to move beyond the regular stateless container + database model. Batch data processing can be solved using the map-reduce application model, and predefined values can be computed with streams. But how can we compute over large data sets at serving time?
This talk will introduce Vespa – a platform for serving-time computation over large data sets. Vespa was recently released as open source from Yahoo – who also originally released Hadoop. We’ll explain why we rely on Vespa for serving in our big data applications involving users and data sets measured in billions, what problems it solves, and how to get started using it.