High Availability InfluxDB 2.0 OSS

Dear community,

I have been using for a while TIG stack (telegraf, influxdb and grafana) in development environment, with all the components running in the same server. I am more than happy with the global performance of the system, nevertheless, it is time to think about industrialization and setting up the production environment…meaning, high availability. I have read a lot of information regarding InfluxDB relay and I have clear that I will need 1 server as load balancer and 2 servers with InfluxDB. However, a lot of doubts appear regarding the infraestructure needed for Telegraf as writing component and Grafana as reading component when InfluxDB will be installed in High Availability mode.

  • Would it be possible to have Telegraf in the same servers as InfluxDB or is better to have another dedicated server just for telegraf?
  • Would it be possible to have Grafana in the same servers as InfluxDB or it needs to be installed in a separated server?
    Regarding Grafana, I have also read some information about HA, which basically consists on Load balancer + serveral instances + shared database between instances.

I would say that all the doubts are fully related to the physical servers needed to set up the HA environment.

Any idea or suggestion regarding the infraestructure for HA will be more than welcome.

Thank you very much.