Forgot The Third Copy

A teammate sizes the storage for a clickstream warehouse with this script and takes the printed number to the capacity meeting.

EVENTS_PER_DAY = 2_400_000_000
BYTES_PER_EVENT = 380
RETENTION_DAYS = 90

total_bytes = EVENTS_PER_DAY * BYTES_PER_EVENT * RETENTION_DAYS
print(round(total_bytes / 1e12, 1), "TB")
# prints: 82.1 TB

The cluster is a three-replica quorum store. Six weeks after provisioning 90 TB the cluster is full. What did the script leave out, and what is the real figure?

A

Compression. Columnar storage would shrink 82.1 TB to roughly 20 TB, so the provisioning was too generous, not too small.

B

The replication factor. Three copies of 82.1 TB is roughly 246 TB.

C

The unit. Dividing by 101210^{12} gives decimal TB, but the disks are sold in TiB, so the true figure is about 74.7 TiB.

D

Nothing is missing. 82.1 TB is correct and the cluster filled because retention was never enforced.