BigQuery Graph queries no longer require an Enterprise reservation
Running a GQL query against a BigQuery graph used to mean provisioning an Enterprise or Enterprise Plus capacity reservation first. Google now lets on-demand pricing cover it too.
Google’s BigQuery release notes for 6 November add on-demand pricing as a way to run Graph Query Language (GQL) queries. Until now, the only route to GQL was a reservation on the Enterprise or Enterprise Plus edition. BigQuery’s own docs now read: “To run GQL queries, you must use on-demand pricing or have a reservation that uses the Enterprise or Enterprise Plus edition.”
BigQuery Graph, which Google introduced in preview in April, lets you model data as nodes and edges and query it with GQL, the ISO-standard graph query language, interoperable with regular SQL in the same statement:
GRAPH graph_db.FinGraph
MATCH (person:Person {name: "Jacob"})-[own:Owns]->(account:Account)-[repay:Repays]->(loan:Loan)
RETURN account.id AS account_id, loan.id AS loan_idWhy it matters: if you wanted to try graph queries on data already sitting in BigQuery, you previously had to commit to a capacity reservation just to test the idea. Now it runs on the same pay-per-byte-scanned billing as any other BigQuery query, which makes a one-off graph query or a small pilot far cheaper to justify.
The caveat: the release note doesn’t say whether on-demand GQL queries carry a different cost profile than SQL queries of similar size, so test on a small dataset before pointing it at anything large.