Examkin

Question

Difficulty: HardDatabase Migration and Schema Conversion using DMS and SCT

A solutions architect is designing the migration of an on-premises 12 TB Oracle database to an Amazon Aurora PostgreSQL-Compatible Edition DB cluster. The source database supports a critical transactional application and experiences high write activity. The database contains several tables with Large Object (LOB) columns that store binary data up to 50 MB, alongside standard transactional tables. The migration strategy requires schema conversion and continuous replication to minimize downtime during the cutover window.

During initial testing of the AWS Database Migration Service (DMS) Change Data Capture (CDC) replication task, the solutions architect observes significant replication lag and CPU spikes on both the source database server and the AWS DMS replication instance. Additionally, some updates are not being replicated to the target database.

Which combination of actions will resolve these performance and replication issues while maintaining the target replication SLA? (Select two.)

  1. Configure the AWS DMS replication task to use the AWS DMS Binary Reader to read the redo logs directly from the source Oracle database, and verify that supplemental logging is enabled at both the database level and table level for all replicated tables.Answer
  2. Separate the tables containing large LOB columns into a dedicated AWS DMS task configured with Full LOB mode, while configuring the main task for all other tables to use Limited LOB mode with a maximum LOB size threshold.Answer
  3. C
    Set the AWS DMS replication task to use Oracle LogMiner for high-performance log analysis, and disable supplemental logging to reduce the CPU overhead on the source database.
  4. D
    Enable supplemental logging at the table level only for tables with LOBs, and run a single AWS DMS replication task with Full LOB mode enabled for all tables to ensure transaction consistency.
  5. E
    Use the AWS Schema Conversion Tool (SCT) to automatically convert all LOB columns to PostgreSQL JSONB data types, and run the replication task without enabling ARCHIVELOG mode on the source database.

Answer

Configure the AWS DMS replication task to use the AWS DMS Binary Reader to read the redo logs directly from the source Oracle database, and verify that supplemental logging is enabled at both the database level and table level for all replicated tables. Also, separate the tables containing large LOB columns into a dedicated AWS DMS task configured with Full LOB mode, while configuring the main task for all other tables to use Limited LOB mode with a maximum LOB size threshold.
The correct combination of actions involves using the AWS DMS Binary Reader to reduce CPU overhead on the source Oracle database by reading redo logs directly, while ensuring that supplemental logging is fully enabled at the database and table levels to capture all changes. Additionally, separating the large LOB tables into their own task utilizing Full LOB mode prevents the main replication task (configured with Limited LOB mode) from being bottlenecked by the slow, row-by-row LOB lookup operations.

Step-by-Step Solution

1
Address source database resource consumption during Change Data Capture (CDC).
Configured AWS DMS to use Binary Reader instead of Oracle LogMiner.
Binary Reader reads redo log files directly, reducing the CPU impact on the source Oracle database instance.
2
Ensure all transactional updates are captured by the CDC task.
Enabled supplemental logging at both the database level and table level on the source database.
Without proper supplemental logging, AWS DMS cannot extract the complete data required to construct replication statements, resulting in skipped updates.
3
Optimize replication throughput for tables containing Large Objects (LOBs).
Split the migration into two tasks: one for LOB tables using Full LOB mode, and one for other tables using Limited LOB mode.
Replicating LOBs in Full LOB mode requires querying the source table row-by-row, which slows down the task. Isolating LOB tables prevents them from bottlenecking the main transactional replication flow.

Key Concept

AWS DMS performance tuning and prerequisite configuration for Oracle CDC migration, specifically supplemental logging, Binary Reader vs LogMiner, and LOB handling strategies.
Estimated Time:3m 0s
Rate this question

Topics

Database Migration and Schema Conversion using DMS and SCT

More questions from this topic