Optimizing Flux query performance and memory usage for high-cardinality downsampling tasks

Hi everyone, I am working on optimizing a set of data processing and downsampling tasks in InfluxDB and running into a performance bottleneck when handling high-cardinality time-series data. Current Setup: InfluxDB version managing high-frequency sensor and metric streams. Running scheduled Flux tasks to aggregate raw data into hourly and daily rollups.