Hello all:

I am wondering if anyone has a remote table strategy such that “hot data” is stored in a table that is backed by RAM.

I have a table with approximately 1 billion events that grows by approximately 1,000 events a minute.

Table is a merge table of merge tables, the parent merge table has a child merge table for every year of data under those child tables are source tables that make up data for areas.

So

Area1_2017, Area2_2017 -> Events_2017

Area1_2018, Area2_2018 -> Events_2018

Events_2017, Events_2018 -> Events

I am finding that as I add more data the system seems to slow down.

I am wanting to keep a cyclical table of say the last 6 months from the Events table in a table backed solely by RAM

So a query to Recent_Events would remain very fast or at the very least constant.

This should be achievable with tmpfs.

I am currently running 11.29.3 on CentOS 7 on VMware 6.7

Server has 128GB of RAM and 16 cores.

Adam