Topic: HyperLogLog and big data analytics for Postgres
Host and pizza sponsor: Neustar
Timon Karnezos will give a talk about the 'hll' extension to Postgres. He'll cover the origin and history of sketching algorithms, a few modern sketching data structures, along with common use cases and implementation guidelines. Finally, he'll cover the guts of the extension and how Neustar is using HyperLogLog sketches to support reporting workloads over hundreds of TBs of unaggregated data.
Timon says "As background, I'll give an overview of what sketching algorithms are, with examples (like Bloom Filters, Count Min Sketch, HyperLogLog, and K-Min Values), use cases, and high-level implementation ideas. Then we'll talk about the PG HLL extension and how we use it at Neustar."
Live video stream will be here. (assuming it works)