What is batch processing?

Batch processing means collecting data over a time window, then processing it as one chunk on a schedule (hourly, daily, weekly).

00:00 -------- events accumulate -------- 01:00
                                              |
                                         batch job
                                              |
                                         table updated

Common shapes

• Nightly warehouse load of yesterday's orders
• Hourly Spark job compacting Kafka dumps into Parquet
• Weekly backfill of a corrected metric

Why teams still love batch

• Simpler ops than streaming (clear start/end, easy to reason about)
• Cheap at huge volumes (process once per window)
• Natural for reconciliations and financial close

Cost of batch

Latency. Dashboards may be hours stale. That is fine for many BI use cases and wrong for fraud alerts.

Interview tip: "Batch = process a closed window of data on a schedule." Contrast with streaming only after you define the latency need.