Not the OP, but if you're looking to drop Heroku (because of cost, or reliability?), it's not that hard to host your own on a VPS or dedicated servers.
Setting it up doesn't look too hard, and you're not going to need it until you reach significant scale - so probably 90% of the people reading this without a dedicated ops team aren't ever going to need it. Startups often seem to over-engineer servers at an early stage or assume they'll need a massive cluster when a single server with multiple processes could serve perfectly well till you actually hit some sort of scaling problem (i.e. millions of users a month). As an example, the HN website ran for a long time on one server (not sure what they use now, but it's not heroku :). Very large VPS instances often cost less than scaling with heroku.
Also, other providers like Linode offer load balancers without setup for a low cost.
What do you do when one of your droplets disappears or crashes?
ansible myserverplaybook.yml
to set up another one in a few minutes. But you'll find this just isn't necessary unless the server config changes - uptime will be measured in months on most providers (not sure about DO in particular, haven't hosted anything serious there).
Where do you host your database?
On another DO droplet, with remote backups, or even on the same one if you have a sizeable VPS and low to moderate traffic.
Set up your own HAProxy instance?
https://www.digitalocean.com/community/tutorials/how-to-use-...
Setting it up doesn't look too hard, and you're not going to need it until you reach significant scale - so probably 90% of the people reading this without a dedicated ops team aren't ever going to need it. Startups often seem to over-engineer servers at an early stage or assume they'll need a massive cluster when a single server with multiple processes could serve perfectly well till you actually hit some sort of scaling problem (i.e. millions of users a month). As an example, the HN website ran for a long time on one server (not sure what they use now, but it's not heroku :). Very large VPS instances often cost less than scaling with heroku.
Also, other providers like Linode offer load balancers without setup for a low cost.
What do you do when one of your droplets disappears or crashes?
ansible myserverplaybook.yml
to set up another one in a few minutes. But you'll find this just isn't necessary unless the server config changes - uptime will be measured in months on most providers (not sure about DO in particular, haven't hosted anything serious there).
Where do you host your database?
On another DO droplet, with remote backups, or even on the same one if you have a sizeable VPS and low to moderate traffic.