A bit about cacti DB and other tuning possibilities that might help in larger setups.
First of all, if i am talking about a larger setup, i am talking about several remote poller systems, thousands of hosts and hundretthousands of data sources. This is the setup used in a datacenter of a company i am working for (in 2021-04):
- one failover cluster with two servers as master poller, 2x10 core CPUs + HT, 64GB RAM, SSD-backed storage
- several remote systems, usually VMs with 4 vCores and 16GB RAM
- about 3500 network devices of different vendors, classes and types (switches, routers, firewalls, load balancers, several other devices)
- 450.000 graphs implemented
- 900.000 data sources
The following chapters describe the components and their performance considerarions.
- The Basics is describing a few things inside cacti, to understand the tuning considerations.
- Hardware considerations gives you some ideas to plan your hardware setup.
- Database Optimizations shall provide an overview about the database tuning.
- Polling Optimizations provides ideas about how to solve specivic problems that are related to high polling time issues.