Presented by:

3151668c5dd4465e37f349c0489d57ee

Dan Wright

Everactive

I have been interested in data and data systems for many years across a variety of disciplines, including scientific high-performance computing, bioinformatics, and music recommender systems. Currently I work at Everactive on our predictive-maintenance cloud data platform for industrial IoT. I'm also real into bikes.

No video of the event yet, sorry!

I will discuss the migration of our industrial IoT data (and ingest pipeline) from OpenTSDB to Postgres + Timescale. When Everactive was first getting started, we didn't understand timeseries databases well, but we knew that timeseries in SQL was well known to be a Bad Thing. As we've grown, we've discovered much about dedicated timeseries stores (and their shortcomings), and Timescale has come on the scene to address the pain points of IoT sensor data in Postgres. You'll hear about the pain points that led to a crisis with our previous timeseries DB, the complexities and questions we had to answer in order to migrate to Timescale, and how having an easily intelligible SQL model for sensor data has both improved platform performance and greatly enhanced productivity for our data scientists and developers.

Date:
2020 March 26 12:40
Duration:
20 min
Room:
Carnegie/Alvin
Conference:
Postgres Conference 2020
Language:
Track:
Migration
Difficulty:
Medium