Storage
RqLite is used as a storage layer. Read more about partitions in cluster section
RqLite is a distributed relational database that combines the simplicity of SQLite with the robustness of a fault-tolerant, highly available system. It's developer-friendly, exceptionally easy to operate, and it's designed for reliability with minimal complexity.
We use the most important parts of the RqLite project as our storage solution. RqLite documentation is a great resource to see most of the configuration possibilities.
RqLite Change Data Capture can be enabled with cluster.cdc.enabled and
cluster.cdc.output, or with RQLITE_CDC_ENABLED and RQLITE_CDC_OUTPUT.
Set the required stable source identifier with cluster.cdc.serviceId,
RQLITE_CDC_SERVICE_ID, or service_id in an advanced output settings file.
The output accepts an HTTP(S) endpoint, stdout, or a path to an advanced JSON
output settings file. CDC exports changes only; it
does not create an initial snapshot or backfill existing rows. Every voting
replica must use the same endpoint and compatible CDC settings. See the
configuration reference for
service identifiers, safe endpoint cutover, setup, and delivery semantics.
We plan to expose the RqLite endpoints on the system path so that the users can use those endpoints to interact with the database directly.