The Distributed Database Internals of InfluxDB

About the talk

In this talk we'll go over the design considerations, technologies, and tradeoffs of InfluxDB, an open source distributed time series, events, and metrics database. We'll talk about how data is replicated within a cluster, how failover occurs, and the evolution of how we decided to split data across a cluster of machines. We'll touch on distributed consensus with Raft, replication fault tolerance with a write ahead log, and how we schedule frequent tasks to run in a reliable way across a cluster.

About the speaker

Paul Dix is the CEO of InfluxDB (YC W13), organizer of NYC Machine Learning, series editor for Addison Wesley's "Data & Analytics", and author of "Service Oriented Design with Ruby and Rails".

For the last sixteen years he has worked as a software developer, tester, network engineer, and even as a PC technician. From late 2005 to 2011 his primary focus was been Ruby on Rails and the Linux server stack.