Optimizing the ATLAS metadata database architecture in preparation for High Luminosity LHC

ATLAS Metadata Interface (AMI) is a generic ecosystem for meta data aggregation, transformation and cataloging, benefiting from more than 25 years of feedback in the LHC context. This paper shows how it is planned to optimize the ATLAS metadata database architecture in preparation for High Luminosity LHC. In particular, the new mechanism for propagating physics parameters along the ATLAS dataset inheritance chain will be presented. This paper also describes the automatic scoping mechanism, that has been introduced in AMI, to take advantage of database partitioning in a transparent way.