We're launching Fly machines today

Example of a golang app using SQLite running on Fly Machines with code.
When no user is present it shuts down. Likewise when a user starts to hit it again, it starts up.

Example of a golang app using SQLite but not on Fly Machines, and so runs all the time.
It has the LiteStream backup and restore feature though.
GitHub - TylerSustare/pocketbase-framework-litestream: A small example using PocketBase as a Go Framework with replication/restoration of SQLite DB with Litestream.

Killer feature will be for the SQLite DB running on Fly Machines to also incorporate backup and Restore.

@benbjohnson @benbjohnson1 Hey dude you have 2 handles, so i chose both.

2 Likes

Author of WunderBase here. We’re launching a managed service soon. Using the machines API directly is just way too complicated. We wrote our own “Provisioner” in Go which wraps the fly GraphQL API and the machines REST API to make the process of managing machines easier.

Adding backups will be one of the first features we’d like to add. Once LiteFS supports backups and replication, we’ll probably switch over.

We will also talk about how we’ve build a CI pipeline with machines that can do a full run in 3s (no change). But we need a Blogpost to get into the details. But in general, fly machines and Podman are a really great match for rootless OCI builds.

Happy to champion fly machines. Excited for “pause and resume” functionality, as it should reduce cold starts further I hope.

3 Likes

I worked on a big-kv-database team at the biggest-cloud-provider, and believe you me, the backup and restore feature came after it had reached gazillion dollars in revenue (:

I’d reckon, one’s for LiteFS (after Fly acq), the other’s for Litestream (before acq) :wink:

1 Like

That’s really interesting to me. I would like to try Wonderbase.

I have very similar needs for an Open Science project where I need to allow each Scientist to run a Fly Machine for their data and logic and then reference it in their Research paper .
This will allow reproducibility of science in terms of how they used the data to come to their conclusions.

1 Like