Using on Disk Storage With an In-Memory Graph Database

Memgraph, an in-memory graph database, has introduced a solution for handling large properties that aren't used in graph computations through its GQLAlchemy library, which provides on-disk storage support. This is achieved by using the SQLitePropertyDatabase to store extensive metadata separately from the graph database to enhance efficiency and manageability. The GQLAlchemy library, built on Pydantic, functions as an Object Graph Mapper (OGM) that connects graph database objects with Python objects, allowing users to define which properties are stored on disk or in memory via model definitions. This approach alleviates the burden of manually implementing separate storage solutions for large properties, thereby improving performance and simplifying project maintenance. Through the use of the on_disk argument, users can automate the storage process, ensuring that large properties are efficiently managed without consuming excessive RAM, thus speeding up graph algorithms.