Optimizing Database Write Performance: Batching vs. Direct Inserts?

Hey everyone, I'm working on a system that experiences high write volumes to a PostgreSQL database, particularly for logging and telemetry data. We're currently doing direct, individual inserts, but I'm wondering if a batching strategy would offer significant performance gains. My main concern is the trade-off with increased latency for individual writes if they have to wait for a batch to fill up. Has anyone implemented a batching mechanism for high-volume writes? What were your experiences regarding throughput improvements and any downsides related to latency or complexity of the batching logic? Are there specific thresholds where batching becomes clearly superior, or best practices for determining batch size and frequency?