DataMosaicDocs · Connecting to data
SQLite and Snowflake
The two drivers that connect by something other than a host and a port.
#SQLite
A file on the machine running Mosaic. No host, no port, no credentials.
| Field | Default | Notes |
|---|---|---|
| Database file | — | "Absolute path to the .db or .sqlite file, on the machine running Mosaic." |
| Table | — | Optional — see SQL databases |
| Open read-only | On |
Open read-only is on by default and the help says why:
A check should not be able to write to the database it reads. Untick only if a query genuinely has to.
The path is on the server's filesystem, not yours. On a hosted installation that means a path inside the container, which is rarely what anyone wants — SQLite is most useful for a self-hosted Mosaic reading a file on the same machine.
#Snowflake
Connects by account identifier rather than by host.
| Field | Notes |
|---|---|
| Account | "The account identifier, such as xy12345.eu-central-1." |
| Username | |
| Password | |
| Warehouse | The compute warehouse to run in |
| Database | |
| Schema | |
| Table | Optional |
| Role | The role to assume |
The fields are in that order in the panel because that is the order Snowflake's own documentation puts them in.
Warehouse is the one to get right for cost. A Check that runs nightly against an oversized warehouse is a recurring bill for work that takes seconds.