DuckDB 2.0 LTS pulls the data stack toward one process
DuckDB 2.0.0 LTS "Cinnamon Teal" shipped. One commenter says it's much faster on high-core EC2 machines, for computation as well as reading from S3, with near 100% utilization. Another says its new ability to run client-server is a game changer that will replace many data stacks running on lots of servers with large CPU and RAM.
Praise is broad. People call it versatile for ingesting and reshaping data, not just analysis. This fits a pattern of simple, single-binary tools eating infrastructure that used to need a cluster, at the same moment a database vendor like PingCAP is in the news for a breach of its cloud control plane.
So what?
Before you spin up a warehouse or a managed distributed database, test whether DuckDB on one big machine handles your workload. It can cut both cost and the number of third parties holding your data.