It offers following features
A simple, high-level, stream-oriented interface
A naming and metadata scheme for managing streams and other entities at a higher layer
A mastership scheme based on ZooKeeper
Log data management policies (e.g. retention and log segmentation policies)
Geo-replicated logs
A fast write pipeline using batching
A fast read pipeline using a read-ahead mechanism
A serving tier that supports write fan-in and efficient read fan-out using caching
https://dzone.com/articles/distributed-logging-architecture-for-microservices