John Swain

We are all familiar with the experience that traditionally, data teams spend a large proportion of their time on data cleansing, preparation, and discovery rather than generating business value. The Medallion Architecture - with its Bronze, Silver, and Gold zones - addresses this through progressive data refinement, and it has become a widely adopted pattern for modern data platforms. However, these implementations create a critical organisational problem: to maintain quality and governance, centralised control is required across all zones, forcing organisations into a false choice between control and agility. Organisations that maintain strict central control preserve data quality but throttle team agility - innovation moves only as fast as data engineering bandwidth allows. While the resulting data is of high quality with standard, well- documented (canonical) models which is critical for AI models, the pace of delivery lags behind business ambition. Those that loosen control to enable speed risk ungoverned proliferation, inconsistent models, and eroded trust in data. Neither approach addresses the fact that a traditional Medallion Architecture can be hindered by its own success when large and / or unpredictable workloads directly affect performance. For business and senior stakeholders, the consequences are tangible: delayed time-to-insight, opaque platform costs, and inability to capitalise on opportunities. Most organisations default to one side or the other, not…

The growth of connected data has made graph databases essential, yet organisations often face a dilemma: choosing between an operational graph for real-time queries or an analytical engine for large-scale processing. This division leads to data silos and complex ETL pipelines, hindering the seamless integration of real-time insights with deep analytics and the ability to ground AI models in factual, enterprise-specific knowledge. Google Cloud aims to solve this with a unified "Graph Fabric," introducing Spanner Graph, which extends Spanner with native support for the ISO standard Graph Query Language (GQL). This session will cover how Google Cloud has developed a Unified Graph Solution with BigQuery and Spanner graphs to serve a full spectrum of graph needs from operational to analytical.