How can I improve my time series ingestion/throughput?

Make sure you’re partitioning and sorting by metric name. Here are some other schema design tips, and here’s some context.