Episode 194: Michael Hunger on Graph Databases

Problem Modeling

Graph databases offer a powerful way to model complex problem domains, especially those with numerous associations. Michael Hunger explains that unlike relational databases, which focus on set-based algebra, graph databases excel in handling connected data structures through nodes and relationships 1. This makes them ideal for applications where data is highly interconnected, such as social networks or recommendation systems. He illustrates this with a movie database example, where elements like movies, actors, and users are nodes, and their interactions form relationships 2.

Graph databases can model complex domains very easily, and it's as they are mostly about the association.

--- Michael Hunger

Choosing the right data store depends on the data model's connectivity and access patterns, making graph databases a superior choice for connection-heavy applications 1.

   

Real-world Examples

Real-world applications of graph databases showcase their versatility and efficiency in handling complex queries. Michael Hunger describes how recommendation algorithms, like collaborative filtering, are simplified with graph databases, allowing for efficient data retrieval and manipulation 3. This capability transforms the design and thought process, focusing on use cases rather than just data structures.

It moves your thinking more around the use cases. So you go away from just thinking about data structures to what you actually want to achieve.

--- Michael Hunger

Additionally, challenges like graph sharding are addressed by focusing on application-level solutions, as seen in Adobe's global Neo4j cluster, which is sharded by geography 4.