In 2025, Patreon focused on maintaining features for its millions of users while overhauling its infrastructure. Their Year in Review highlights 12 projects that emphasize architectural changes, data model refactoring, and consistency trade-offs in distributed systems. Key strategies included defensive migrations, decoupling data relationships, and improving analytics through a new transformation layer.

This article discusses LinkedIn's approach to improving HDFS block placement for its massive data clusters. It explains how they adapted their block placement policy to streamline maintenance operations, reduce data replication, and maintain high data availability. The changes were necessary due to the challenges of managing over 5 exabytes of data efficiently.