I am in the process of choosing data storage for my event-series DB. There are various opinions on using RDMS, NoSQL or Columnar. Was there any exploration done on using Time-series DB for event-sourcing other than TimeScaleDB(https://blog.timescale.com/time-series-data-why-and-how-to-use-a-relational-database-instead-of-nosql-d0cd6975e87c/)? Or is it even a good idea to use Time-Series DB for event sourcing?
I think you need analyze your requirements first, and then let that you point you towards the databases that provide the functionality you need. Then evaluate and compare the databases.<p>I would start by asking if a RDMBS or MPP warehouse can work. If it can't, why not?<p>I run a company that captures billions of events a day (an event source pattern) and we use Snowflake. I chose snowflake because it separates compute from storage (effecively paying $.20/compressed TB/month for storage) and allows me to elastically expand the compute required for analysis or alteration of data.<p>I don't need need most the real time features of a time serieis DB (but I could easily split the data out of Kafka to a TSDB if I needed). I need the ability to alter the data and run sophisticated aggregations across the data.<p>Bottom line: if you're not certain you need a TSDB, or know what features you need yet, I wouldn't go chasing down a TSDB. Let your requirements push you to a database.
I've already look at <a href="http://pumpkindb.org/" rel="nofollow">http://pumpkindb.org/</a> but finally use TrailDB. Else, there's also the leading InfluxDB for Time-Series DB. But if the volume isn't very high, a MySQL/PGSQL could satisfy no?