Here is a AWS Aurora PostgreSQL 12 database.
The database is an r5.4xlarge and average load is ~16.
Over 3 hours, the average "Write IOPS" was 40,000 and the average "Read IOPS" was 15,000.
In the same period, the average "[Billed] Volume Write IOPS" was 400,000 and the average "[Billed] Volume Read IOPS" was 3,000,000. (I believe those are 5 min sums.)
Why are writes 2.5x reads, but for billed volume, writes are 1/8th reads?
Why is there a huge difference? Is that unusual? Does it indicate an I/O pathology?
