For quite some time now I've been sitting on some ideas how Kafka Connect could be made more "Kubernetes-native", for instance achieving better isolation between connectors. So far it's an ideation only, but would folks here be interested in reading about this?
Took a stab at adding #io_uring support to ClickHouse: github.com/ClickHouse/Cli…, seeing ~5-20% performance improvement with reduced resource utilization
@mituzas Figured this would look nicer written in bpftrace: gist.github.com/sauliusvl/018a…, but probably not that much once you add all the filters, also formatting is less customizable (still can't fold stacks into single lines)