For stackql, an SQL backend is an abstraction that executes SQL statements. The workload is is light on for updates and even lighter for read-write contention.

At present, traditional RDBMS systems such as sqlite and postgres are used to implment SQL backends. However, there is no reason that a non-traditional, analytics optimised system can not be used at least in part.

The default implementation is embedded SQLite. SQLite does not have a wire protocol or TCP-native version.

• Using golang SQL driver interfaces.
• PGX native (improved performance).

https://github.com/fergusstrange/embedded-postgres

Postgres failing tests checklist:

• Google IAM Policy Agg
• Google Join Plus String Concatenated Select Expressions
• Okta Users Select Simple Paginated
• AWS EC2 Volumes Select Simple
• GitHub SAML Identities Select GraphQL
• GitHub Repository With Functions Select
• Join GCP Okta Cross Provider JSON Dependent Keyword in Table Name
• K8S Nodes Select Leveraging JSON Path
• Data Flow Sequential Join Select With Functions Github
• Shell Session Simple
• PG Session Postgres Client Typed Queries
• PG Session Postgres Client V2 Typed Queries

How to use drivers:

• https://go.dev/doc/database/open-handle

List of drivers:

• https://github.com/golang/go/wiki/SQLDrivers

• SQLite as per golang (modernc.org/sqlite).
• Postgres URI.