LiteFS Cloud: Distributed SQLite with Managed Backups

Summary

LiteFS Cloud, launched in preview, adds managed backups and restores to LiteFS, an open-source distributed SQLite database that replicates data across application-cluster nodes. Existing LiteFS users had to create snapshots, choose retention, and accept restoration points limited by snapshot frequency; the service instead provides an equivalent snapshot every five minutes and restores within the last 30 days at five-minute granularity. Setup requires LiteFS 0.5.1 or later, a dashboard-created cluster, and an auth token, after which backups are automatic and dashboard restores use the latest snapshot before a selected timestamp. Its LTX format stores changed database pages, while multi-level compactions combine daily snapshots, hourly changes, and five-minute intervals so a restore can stream-merge a minimal set of files. To improve async replication’s durability trade-offs, the primary batches changes, LiteFS Cloud persists them in S3, and a high water mark prevents local transaction files from being removed before storage confirmation.

Takeaways

1. 1

   LiteFS Cloud is in preview and currently free; setup requires LiteFS 0.5.1 or later, a LiteFS Cloud cluster, and an authentication token.

2. 2

   Multi-level LTX compactions use daily full snapshots, hourly change files, and five-minute intervals to minimize the files needed for point-in-time restoration.

3. 3

   The LiteFS primary batches changes every second, while LiteFS Cloud periodically flushes compacted data to S3 and propagates the latest persisted transaction ID back to cluster nodes.