Replicating operational databases into analytical storage is a common practice in data lakes. However, querying these operational systems can strain resources and hinder performance for complex analytics. Change Data Capture (CDC) addresses this issue by capturing real-time changes—like inserts, updates, and deletes—from the source database's transaction log. Instead of having to query directly, CDC tools read from logs, ensuring the operational workload remains unaffected. Several tools facilitate CDC, with Debezium being a popular open-source option that supports multiple database types and outputs. AWS Database Migration Service (DMS) is another choice, particularly in AWS environments, though it offers less flexibility. Flink CDC leverages Apache Flink for a streamlined ingestion process into Iceberg tables. When integrating CDC data into Iceberg, two main strategies emerge: direct materialization, where changes are written directly to the final table, and the raw change log with ETL, which involves writing changes to a bronze table before processing them into a mirror table. The latter allows for more flexibility in transformation and handling of historical data. However, scaling CDC presents challenges. High update workloads can strain system performance, especially when dealing with large tables. The choice between Copy-on-Write and Merge-on-Read significantly influences performance. Copy-on-Write rewrites entire data files for each update, leading to increased latency and resource use. Merge-on-Read, while keeping ingestion fast by writing delete markers instead of rewriting data files, can degrade query performance if not managed properly. Partition layout also plays a critical role; time-based tables are easier to manage than entity tables, which may require updates across multiple partitions, complicating compaction processes.