This document provides an overview of real-time indexing in Druid. It describes the key components of Druid's real-time indexing architecture including Tranquility, the indexing service, firehose, plumber and real-time tasks. Tranquility is used to ingest event streams from Kafka in real-time and submit indexing tasks to Druid. The tasks read data from the firehose, incrementally build indexes, and push completed segments to deep storage via the plumber. The document explains how these components work together to continuously ingest and index streaming data.