Soaring backup and IO utilization with a new Aurora PostgreSQL instance

I recently migrated a 6TB Postgresql database from RDS to Aurora. Within the first week and with what I would consider relatively light activity, I saw my BackupRetentionPeriodStorageUsed peak at 42 TB before trending back down. Similarly, my total billed I/O requests hit around 10 billion units. If I understand correctly, at up to 8kb per unit that's potentially enough to write 80 TB. I don't think my workload amounted to even a tiny fraction of that. Is there a certain amount of overhead that comes with on-boarding to AWS? Or perhaps I'm getting a terrible fill rate on the I/O requests or the cost of maintenance activity like vacuuming is extraordinarily high? Subsequent billing was much closer to my expectations as backup and io was much more subdued. I don't understand. Is this normal or was I over-billed by wonky behavior of the data storage tier?