What changes for you
An unsliceable table still moves.
If range discovery cannot create reliable boundaries, Omni Loader warns and falls back to one unsliced work item instead of inventing unsafe predicates.
Scale · Table slicing
Turn one very large table into parallel, bounded work.
Omni Loader divides large tables into independent slices using source partitions or discovered ranges, then schedules the biggest work first to keep workers productive.
Parallel work planner Divide · schedule · balance
Workers active
Inputs
PartitionsREADY
Date + number rangesREADY
Text + file rangesREADY
Slice planner Divide · schedule · balance
Outputs
Worker 01READY
Worker 02READY
Worker 03READY
Why this path matters
Use structure the source can expose.
Create slices from native source partitions, numeric ranges, dates and times, text prefixes, an explicit slice column, or file row ranges.
- ✓Source partitions
- ✓Numeric and temporal ranges
- ✓Text-prefix ranges
- ✓File row ranges
The payoff
See what the capability changes.
1–64
Worker control
Set the concurrency budget for the migration environment.
1–3,600
Slices per table
Break exceptional tables into manageable ranges.
Largest first
Useful scheduling
Keep workers occupied instead of leaving the hardest table until last.
Where the work happens
The details that make the result usable.
At a glance
A visual summary of the work.
Separate ingestion capacity
Balance extraction and destination work.
Source/target workers and ingestion workers have separate controls so staged cloud loads can be tuned around both extraction and target ingestion.
Take the next step
Give the largest tables a migration plan of their own.
We will identify usable slice keys, source pressure, worker limits, and the target ingestion budget.
Plan my migration