That's good news, especially the reduced memory usage and OTLP ingestion support look nice. I have experimented with OTLP metrics before but eventually fell back to prometheus to avoid adding another service to our systems.