Digital transformation initiatives have unlocked large and fast-moving data sets including clickstreams, network telemetry, application monitoring and IoT devices. Analytics architectures have not kept pace, with most #data still being run through existing “cold analytics” systems and tools designed for smaller and less time-sensitive workloads. “Hot analytics” denotes workloads where the responsiveness of the system is instantaneous and can support self-service data exploration, and where the data is extremely fresh, allowing for more informed decision-making.
The breadth of analytical systems in the world today demands a clear approach to selecting the right one for a given workload. In this talk, we’ll discuss a temperature-based way of thinking, where workloads get “hotter” as they become more interactive, more concurrent, and more likely to need up-to-the-second data.
Apache Druid is a modern cloud-native, stream-native, analytics database designed for workflows where fast queries and instant ingest are important. Druid excels at instant data visibility, ad-hoc queries, operational analytics, and handling high concurrency. It is a strong candidate for being the workhorse system for hot analytics.