Thinking in events: from databases to distributed collaboration software

In this keynote I give a subjective but systematic overview of the landscape of distributed event-based systems, with an emphasis on two areas I have worked on over the last decade: large-scale stream processing with Apache Kafka and associated tools, and real-time collaboration software in the style of Google Docs. While these may seem at first glance to be very different topics, there are also important points of overlap. This paper lays out a taxonomy of event-based systems that shows where their commonalities and differences lie. It also highlights some of the key trade-offs that arise in the implementation of event-based systems, drawing both from distributed systems theory and from experience of their practical deployment. Finally, the paper outlines a number of open research problems in this field.