How do you ensure fault tolerance during large-scale data migrations?

Section 1 — The Context (The 'Why')
Fault-tolerant data migration must handle large volumes, schema mapping, and cutover with minimal downtime. The primary challenge is consistency during dual-write and validation. Big-bang cutover risks data loss when validation is skipped.

Section 2 — The Diagram

[Source] --> [Extract] --> [Transform] --> [Load]
   |              |              |                |
   v              v              v                v
[Checkpoint] [Schema Map]  [Validate]    [Dual-Write]

Section 3 — Component Logic
Checkpointing at each stage enables resume from the last successful point after failure. Schema mapping handles type and structural differences between source and target. Validation compares record counts and checksums between source and target; discrepancies trigger alerts. Dual-write during cutover allows gradual traffic shift. Phased rollout with a documented rollback plan limits blast radius. Idempotent loads prevent duplicate records on retry. In production, monitor consumer lag, checkpoint success rate, and sink write latency as primary SLOs. Partitioning strategies should align with query patterns; bucketing within partitions mitigates join skew. TTL policies on raw and intermediate data control storage cost while preserving replay capability for debugging and backfill. Data skew mitigation via salting or secondary hashing prevents single partitions from becoming bottlenecks. Exactly-once semantics require transactional commits at the sink; at-least-once delivery demands idempotent write logic to avoid duplicates. Fan-out patterns allow one source topic to feed multiple downstream consumers without re-ingestion. Backpressure handling ensures that slow processors do not cause unbounded buffer growth; Kafka consumer lag is a key metric. Schema evolution should follow additive-only rules where possible to avoid breaking consumer compatibility. The CAP trade-off should be documented per component: analytics typically favors AP, while financial reconciliation requires CP. Blast radius from component failure is bounded by replication and checkpointing; design for graceful degradation during partial outages. Cost optimization: use Spot instances for batch workloads and tier cold data to lower storage classes. Dead-letter queues preserve failed records for replay rather than dropping them.

Section 4 — The Trade-offs (The 'Senior' part)