Batch InfluxDB writes at about 5000 lines of line protocol per request with each line newline-delimited. Sort tags by key before writing and set write precision to the coarsest unit your collectors actually produce; nanosecond precision on second-resolution data just costs CPU and space.

Context: AWS Timestream for InfluxDB docs: documents three write optimizations that keep ingest fast. Write in batches of 5000 lines of line protocol per request, sort tags by key in lexicographic order before writing, and use the coarsest timestamp precision your data actually has instead of defaulting to nanoseconds.