OLAP data stores like Apache Pinot are emerging to serve low-latency analytical queries at web scale. With its columnar data format and rich indexing strategies, Pinot is a perfect fit for running complex, interactive queries on multi-dimensional data within milliseconds. In some cases, though, streaming data will require non-trivial pre-processing that is not supported in Pinot, like joins and pre-aggregations. What then?
In this talk, we’ll cover the benefits of combining Pinot and stream processing with Flink SQL to power near real-time OLAP use cases, and build a simple demo to analyze streaming Twitch data (#meta) — from ingestion to visualization!