Ramkumar took us behind the scenes of the famed Bloomberg terminal. He described some of the daunting challenges involved in making vast quantities of new content searchable within 100 milliseconds of being published, while also being able to search across exabytes of data at any given time, all under extreme uptime requirements. He then described some of the Solr techniques they are using to achieve these and other feats such as supporting very long and complex queries, delivering notifications on saved searches, and much more.
Mark Harwood then shared some of the new features present in Elasticsearch 1.0. He briefly touched on smarter backup processes and a distributed percolator for things like alerting. But the feature he demonstrated with most fanfare was what Elasticsearch call aggregations — aptly described as facets on steroids. While facets are computed for a single dimension — a list of “genres” for instance — aggregations can be computed for multiple dimensions. A given genre — “thriller” lets say — could contain computations not just for how many films were in that genre, but also what the total, average, minimum, and maximum box office earnings for those films were, and the geographic locations in which they were produced. An earlier version of Mark’s slides can be found here.
The meetup has grown from nothing to over 500 members during the course of four years, with 18 meetups along the way.