HOT AND COLD TIERS
Keep recent rows fast. Park the rest on object storage.
ColdFront moves aging rows into Apache Iceberg files on any S3-compatible store while the table stays a single table. Your queries, models and dashboards do not change.
STEP 1
Choose what goes cold
Move a whole table, or let age decide. The table keeps its name and its columns.
- Tier by age or on demand
- Nothing to rewrite in your queries
Tiers: a table moved to cold storage1400 x 1000 pxDelete this block and drop in an Image or Video widget
Tiers: the same query before and after1400 x 1000 pxDelete this block and drop in an Image or Video widget
STEP 2
Query hot and cold together
A query returns identical results before and after the move. Recent rows come from Postgres and older rows come from Iceberg files.
- Same SQL on both tiers
- Results match on both sides of the flip
STEP 3
Train on cold data
Models can train on data that lives in object storage. A model trained on cold data scored an ROC AUC of 0.94 in the demo scenario.
- No copy back into the database first
Tiers: a model trained on cold data1400 x 1000 pxDelete this block and drop in an Image or Video widget
Tiers: moving a table back to hot1400 x 1000 pxDelete this block and drop in an Image or Video widget
STEP 4
Flip it back any time
Bring a table back to hot storage when it becomes busy again. The move is reversible.