While building the Azure filtered by OpenShift tables in python we use a contains logic to match the resources:

• https://github.com/project-koku/koku/blob/main/koku/masu/util/azure/common.py#L188-L191

However, when we get into resource matching in the Trino SQL, we use an exact match here:

• https://github.com/project-koku/koku/blob/main/koku/masu/database/trino_sql/reporting_ocpazurecostlineitem_daily_summary.sql#L370-L373

For some customers this results from matching 98% of their storage cost in the parquet files, to only matching about 5% of their storage cost in the Trino SQL.