The Avoidable Outage
A database that fills its disk stops accepting writes and can corrupt or crash. Unlike a sudden spike, storage growth is gradual and predictable, which makes running out of space one of the most preventable outages there is. Capacity planning turns that predictability into lead time.
What Drives Growth
Projecting storage needs more than current size:
- Data growth rate: gigabytes added per day from new rows and larger payloads.
- Indexes and overhead: indexes, WAL, and bloat often add more than the raw row data.
- Retention: how long you keep history before archiving or deleting bounds the total.
- Temporary spikes: backups, vacuum, and large migrations need transient headroom.
Plan With Headroom And Lead Time
Estimate the date you would hit a danger threshold, often 75 or 80 percent full, at the current growth rate. Plan the expansion or cleanup to land well before then, because resizing volumes, archiving data, or sharding all take time. Headroom absorbs forecasting error and short spikes.
Key idea
Storage capacity planning projects predictable data growth including indexes and retention, then schedules expansion or cleanup with headroom and lead time so the database never runs out of disk.