[armis]

I've built some dedicated analytics systems handling billions of events per month on the past, here are my comments:

1. Always know your limits. PG is impressive even on a large scale, so I would suggest you clone your existing database server and fill it with dummy data. After this, check how well it performs and when it breaks. Who knows, maybe you can live 5x the traffic with your current setup.

2. I can personally recommend checking HBase[0] and thinking about OLAP[1] based analytics. The relational version of OLAP is very good, but it's not very scalable. However, you can build it on top of a NoSQL solution (like HBase) as I did in the past. It was very robust and performant.

3. Lastly, it doesn't matter which database you select, as long as you understand how it works and know its limits. But generally, you might want to move away from relational databases as they're not designed to deal with massive datasets (although, you always can shard).

[0] https://hbase.apache.org/ [1] https://www.guru99.com/online-analytical-processing.html