Metadata in statistical production is ubiquitous: from concepts, classifications and variables to retention and provenance information, metadata is created, used and shared across all phases of the data lifecycle. Unfortunately, metadata management is sometimes silo-based and tool-specific, which impairs all four FAIR principles (Findable, Accessible, Interoperable, Reusable). Statistics Canada is in the early stages of implementing a virtual metadata integration platform, the Metadata Hub, that integrates a collection of metadata repositories, e.g. Colectica, Aria, SDMX Istat Toolkit, OpenLink Virtuoso and CKAN across a number of standards, e.g. DDI, SDMX, XKOS, DCAT, and RDF/OWL, among others. In this presentation we describe our experience so far and the way forward