Fly Workers and DAAS

Curious to hear where fly workers and database as a service is on the roadmap.


Workers are just around the corner! We’ve been using a worker-like setup ourselves on our own platform for a few months now (running a Buildkite agent). We’re now formalising the process into something that fits a worker type better.

Databases will be possible too with our release of permanent disks and (soon) private networking.


Awesome to hear!

I saw that and was going to ask if this was something that can be used with databases but it looked like they could possibly loose data in certain circumstances or is this not true and they are suitable for DBs?

There’s always a chance of data loss if we lose the specific hard drive where your volume is (due to a physical failure of some kind) or if the server fails. We’re not currently running this on large network storage arrays, just on dedicated servers with large disks.

Ideally you’d run a database cluster that has some replication / resiliency mechanism. One that can “elect” a leader / primary and where failures result in falling back to a follower for that role. There’s a chance we’ll provide an easy way to do this for popular databases. First we need to eat our own dog food and put some significant usage on our own implementation before we can sell that comfortably.

Makes perfect sense, I actually shared something here last night that has this concept but I am sure it’s far from popular. (Dgraph)

I think something like mongo atlas on a fly network could be super cool.

Great to hear workers are coming soon!