Explore a conference talk detailing the migration from PostgreSQL to HDF5 for storing large amounts of floating-point data in scientific projects. Learn about performance issues encountered with PostgreSQL when handling complex multidimensional arrays, and discover how switching to HDF5 file storage solved these problems. Gain insights into migration strategies, managing a mixed environment of metadata in PostgreSQL and data in HDF5, and addressing concurrency in distributed and parallel applications. Understand the process of measuring, monitoring, and testing legacy systems while rewriting code to improve performance without disrupting user experience. Benefit from real-world examples and practical advice on optimizing data storage for scientific applications.