BigQuery Data Clean Rooms vs pgloader

BigQuery Data Clean Rooms let multiple parties share, join, and analyze data without moving or revealing the underlying data. The cloud service works with BigQuery sharing, and shared resources must be BigQuery tables, views, or table-valued functions. Owners manage rooms, contributors publish data, and subscribers query it. Contributors can apply analysis rules, including aggregation thresholds, and owners or publishers can offer predefined query templates without exposing the underlying tables or views. Subscribers are automatically prevented from copying or exporting raw data, while contributors can configure controls for query results. These controls have limits: Google cautions that analysis rules may not stop sophisticated queries from extracting unauthorized data, and notes that rules can be set only on views. Each room can contain at most 100 shared resources. Availability is restricted to BigQuery sharing regions, without listings across multiple regions. Rooms can be managed in the Google Cloud console or created through the API; setup requires enabling the Analytics Hub API and assigning required IAM roles. Contributors pay for storage, while subscribers pay for compute when running queries.

pgloader migrates database and flat-file data into PostgreSQL using schema translation, casting rules, and parallel loading. Listed sources are MySQL, MariaDB, MS SQL Server, SQLite, CSV, fixed-width files, and DBF; Oracle support is not built yet. Its pipeline fetches the source catalog, prepares the PostgreSQL target, runs parallel COPY, creates indexes in parallel, and completes the schema. MySQL handling includes enum translation, converting TINYINT(1) to boolean, zero-date handling, and looking up credentials in .my.cnf. The MS SQL Server path includes Azure AD authentication, sequence migration, and multi-column foreign keys. For file imports, rejected rows can be separated after a failed COPY transaction while accepted rows continue loading. Custom casting rules go in a .load file. Version 3 in Common Lisp is the stable packaged release; version 4 is a pre-release JAR requiring Java 21 or later. The free Community tier has best-effort GitHub support. Paid support starts at €100/mo, with higher tiers offering priority handling, SLA-triaged patches, roadmap input, or issue-based support. It is self-hosted on Linux.