Choosing datastores

Druid arguably picks design decisions from Timeseries DBs, Datawarehouses and Search systems. In this talk we show why that is so and delve upon use cases that best fit for druid and more importantly when not to choose it. Druid has been the production workhorse for the past 3+ years at Zeotap powering the core Audience planning across our products. Beyond the traditional interactive analytics use case we have used Druid to store non-temporal data (no timestamp) and derive insights. We will go over one data model on how to achieve this. Lastly, we would share our production experiences on operating Druid on GCP cloud.