Skip to content
datastudyguides

DP-750 · 3.2.4 · Prepare and process data

CDC ingestion

Ingest data by using a change data capture (CDC) feed

Apply inserts, updates, and deletes from a change feed in event order instead of repeatedly reloading full snapshots.

See what a change feed contains

Microsoft Learn example showing one changed source row becoming a CDC update record with an operation and sequence number.

A CDC feed carries only inserts, updates, and deletes, plus metadata such as operation type and a sequencing value.

Apply changes in logical order

  1. 1Initial hydrationLoad current source state
  2. 2Stable keyIdentify the target entity
  3. 3Change eventsCapture inserts, updates, deletes
  4. 4SequenceOrder events correctly
  5. 5TargetApply the resulting state
CDC processing

Arrival order is not always business order. AUTO CDC can use a sequencing column to handle out-of-order events.

CDC versus snapshots

  • CDC feed

    Contains
    Only changes
    Best API
    AUTO CDC
    Efficiency
    Incremental

  • Snapshot

    Contains
    Full table state
    Best API
    AUTO CDC FROM SNAPSHOT
    Efficiency
    Compare versions
Source change format

Use real CDC when the source exposes it instead of repeatedly reloading complete snapshots.

Key points

  • CDC feed: Carries row-level source changes
  • Keys: Identify records that changes belong to
  • Sequence: Defines correct event order
  • AUTO CDC: Applies changes to streaming targets declaratively

Exam trap

CDC needs reliable keys and sequencing; arrival order alone may not equal business change order.

Sources · 2 Microsoft Learn pages Show ▾

Checked on October 4, 2026. Report a problem

Go deeper on Microsoft Learn

Checked against Microsoft Learn on October 4, 2026. Report a problem with this card

How well do you know this?