Not the whole db. Only data related to the day itself. A duckdb file is cached generated from a restful url (e.g /api/v1/risk/date/scheme) so that resides in

Not the whole db. Only data related to the day itself. A duckdb file is cached generated from a restful url (e.g /api/v1/risk/date/scheme) so that resides in a risk-scheme-date-uuid.duckdb) then a query will find the latest risk-scheme-date-*.duckdb and based on certain filters return a subset of the data. It’s more complex than that of course as the db is opened in read only to allow concurrent reads, a set of tmp tables are created on each read request to represent user rights where we filter returned data based on what the user is allowed to see. Each duckdb file is around 50mb and usually a newer file of matching scheme is written every 10 minutes, then the old ones are purged when they’re older than two hours