A DynamoDB table in provisioned mode has target-tracking auto scaling enabled. A batch job jumps from low traffic to a large write burst within seconds, causing temporary throttling before capacity increases. Which operational change best addresses repeated scheduled bursts without depending on auto scaling reacting instantly?

1. Lower the maximum write-capacity setting so the target-tracking policy reacts more aggressively.
2. Increase provisioned write capacity before the known batch window and return to the normal configuration afterward.
3. Disable exponential backoff in the batch client so retries reach DynamoDB immediately.
4. Reduce the number of partitions by assigning every item the same partition-key value.