I am running my discourse on a single droplet on DO and I would like to create a more redundant setup. All the resources I found online points me in the direction of load balancer but for doing so I need to create an external DB droplet, move the DB from prod and so on.
Does anyone have experience on something similar?
I imagine many providers have some level of implicit failover. Have you confirmed with DO what level of failover they provide with a droplet?
As an aside I’ve been running two forums on Scaleway for several years and had maybe 30 mins of downtime not caused by a deliberate rebuild? Can only imagine DO has similar robustness. Why do you need to improve on that kind of SLA? That’s a better level of robustness than most banking systems I’ve known!
If you want (for some reason) a fully HA setup in Digital Ocean I’d do something like:
Droplets with the application container
Managed PostgreSQL Database from DO
Here’s some information about the configuration for the external DB.
This will increase greatly the monthly bill for the infrastructure, and I really think it’s unnecessary. Also, this will not protect you from a massive data center issue as Digital Ocean doesn’t have Multi Availability Zones per region, or inter region load balancing (at least that’s my understanding).
In the five years or so I’ve been using Digital Ocean, I haven’t seen any issue that would affect in such a way the Data Center.