Multitenancy shared vs seperate schema

I never used Triplex, so I can’t say anything on that, but Ecto has a few guides on handling Multi tenancy:

The company I work at uses FK on all schemas similar to the second guide of Ecto (shared schema). However we never implemented the prepare_query/3 to add the tenant id to all queries of a repo, since we didn’t like the “magic” of it. Instead we just pass the tenant struct to each function that controls anything related to a tenant (e.g. list_users(%Tenant{id: tenant_id}), do: %User{} |> where([u], u.tenant_id == ^tenant_id) |> Repo.all()). It is a bit cumbersome but does provide flexibility and is a bit more explicit on what’s happening. It just comes down to personal taste I guess.

1 Like