Deployment and Migration practice questions

Install SQL Server Agent on an Amazon EC2 instance and connect to RDS.

Why wrong: SQL Server Agent cannot connect to RDS for SQL Server from an EC2 instance; RDS does not support SQL Server Agent.

Use the AWS Schema Conversion Tool (AWS SCT) to convert the jobs to PostgreSQL functions.

Why wrong: AWS SCT is for schema conversion, not for job migration, and the target is SQL Server, not PostgreSQL.

Use AWS Database Migration Service (AWS DMS) tasks to replicate the jobs.

Why wrong: AWS DMS does not migrate SQL Server Agent jobs; it only migrates data.

Deploy with one or more read replicas in the same region

Why wrong: Read replicas use asynchronous replication, which can introduce read-after-write inconsistency. They are not suitable when strong consistency is required.

Deploy with cross-region replicas

Why wrong: Cross-region replicas have higher latency and also use asynchronous replication, so they do not guarantee read-after-write consistency.

Deploy as Multi-AZ with a synchronous standby

Why wrong: Multi-AZ provides synchronous replication for high availability, but the standby is not used for reads and synchronous replication adds write latency, which may not be necessary for the given requirements.

Export data using mysqldump and import into Aurora during a maintenance window.

Why wrong: mysqldump requires downtime and is not minimal downtime.

Use AWS Schema Conversion Tool (SCT) to convert the schema to Aurora MySQL.

Why wrong: Schema Conversion Tool is not needed when migrating from MySQL to Aurora MySQL.

Use AWS Database Migration Service (DMS) with change data capture (CDC) to migrate data.

Why wrong: DMS with CDC is a valid method, but the question expects a different combination.

The RDS instance is retained and the stack deletion fails.

Why wrong: This would only happen if DeletionProtection or a 'Retain' DeletionPolicy were explicitly set. The stem does not indicate any such configuration.

The RDS instance is deleted without a final snapshot.

Why wrong: Incorrect. By default, a final snapshot is taken. Skipping it requires setting SkipFinalSnapshot to true.

The RDS instance is retained and a final snapshot is taken automatically.

Why wrong: Incorrect. The instance is deleted, not retained; the default DeletionPolicy is 'Delete'.

AWS Trusted Advisor

Why wrong: Trusted Advisor does not assess database compatibility.

Amazon CloudWatch

Why wrong: CloudWatch monitors performance, not compatibility.

Amazon Inspector

Why wrong: Inspector is for security vulnerability assessment.

Use a custom application to capture MongoDB oplog changes and apply to DocumentDB.

Why wrong: A custom application capturing MongoDB 4.0 oplog changes would fail because DocumentDB 3.6 does not natively support the oplog format or commands introduced in MongoDB 4.0, leading to errors. This approach is tempting as oplog capture is a standard method for online migrations with minimal downtime, and would be effective if the source MongoDB version were compatible with the target DocumentDB version, such as MongoDB 3.6 to DocumentDB 3.6.

Configure DocumentDB as a replica of the MongoDB primary.

Why wrong: Incorrect. DocumentDB cannot be configured as a replica of a MongoDB primary; they are separate systems with different replication mechanisms.

Export MongoDB collection data to JSON files, then import into DocumentDB.

Why wrong: Incorrect. Exporting to JSON and importing is an offline approach, not suitable for minimal downtime migration.

Set the source database recovery model to simple to reduce log size.

Why wrong: Setting the recovery model to simple reduces log size but disables point-in-time recovery and CDC, making it unsuitable for minimal downtime migration.

Use AWS SCT to convert the schema to Amazon Aurora.

Why wrong: AWS SCT is used for converting schema to another database engine (e.g., Aurora), but the target is SQL Server, so this step is not needed.

Enable full supplemental logging on the source database for DMS CDC.

Why wrong: Full supplemental logging is an Oracle-specific requirement; for SQL Server CDC, you must enable Change Data Capture and set the recovery model to full, not this logging mode.

Export the database to a flat file, compress it, and upload to S3 via the internet, then import into RDS.

Why wrong: Over 100 Mbps, transferring 3 TB would take over 3 days, exceeding the deadline.

Use AWS DataSync to transfer the database files directly to Amazon S3.

Why wrong: DataSync is for file transfers, not database migration directly to RDS.

Establish a VPN connection and use AWS DMS with full load only.

Why wrong: VPN does not improve bandwidth; full load alone would take too long.

Store database backup files in Amazon S3

Why wrong: Storing database backup files in Amazon S3 is a step for native backup/restore migration, but it does not minimize downtime as it requires downtime for the restore operation.

Use AWS Schema Conversion Tool to convert the schema

Why wrong: The AWS Schema Conversion Tool (SCT) is used for schema conversion when migrating between different database engines, not for minimizing downtime. Additionally, the source and target are both SQL Server, so no conversion is needed.

After migration, enable Multi-AZ for high availability

Why wrong: Enabling Multi-AZ after migration provides high availability and automatic failover, but it does not reduce downtime during the migration process itself. This action is not relevant to minimizing migration downtime.

Use AWS Schema Conversion Tool (AWS SCT) to convert the schema and AWS DMS for data migration

Why wrong: This approach is valid but adds unnecessary complexity because AWS DMS can perform the schema conversion itself for Oracle to PostgreSQL, making a separate SCT step redundant. Thus, DMS alone is sufficient.

Use pg_dump to export the Oracle database and pg_restore to import into Aurora

Why wrong: pg_dump is a PostgreSQL-native tool and cannot read Oracle data. It does not support cross-engine migration.

Use AWS S3 to store the data and AWS Glue to transform and load into Aurora

Why wrong: Using S3 and Glue adds extra steps and is not a standard approach for database migration. AWS DMS is purpose-built for this task.

Use AWS SCT to convert the schema to Amazon Aurora.

Why wrong: SCT for schema conversion is not directly related to minimal downtime migration.

Take a full backup using pg_dump and import it into RDS during a maintenance window.

Why wrong: This approach requires downtime.

Create an RDS Read Replica of the on-premises database.

Why wrong: An RDS Read Replica cannot be created from an on-premises database because Read Replicas are an Amazon RDS feature that replicates only from a source RDS instance, not from an external server. This option is tempting because Read Replicas are commonly used to offload read traffic with minimal downtime during migrations, but here the source is on-premises, so the replica cannot be established. The correct approach would use AWS Database Migration Service (DMS) with ongoing replication to keep the target in sync while the source remains live.

Set the LOB chunk size to 64 KB in the task settings.

Why wrong: Setting the LOB chunk size to 64 KB may reduce memory consumption per LOB column but is unlikely to resolve the out-of-memory error because the total memory on the instance is still 4 GB; increasing instance size is more effective.

Split the 'Orders' table into multiple DMS tasks running in parallel.

Why wrong: Splitting the table into multiple DMS tasks does not reduce memory usage per task; it could increase overall memory consumption and does not solve the memory constraint on the instance.

Change the DMS task settings to disable LOB support for the 'Orders' table.

Why wrong: Disabling LOB support would cause DMS to truncate LOB data to 1024 bytes by default, resulting in data loss, which is unacceptable for a migration requiring minimal downtime.

Use pg_dump and pg_restore to export and import the database.

Why wrong: pg_dump/pg_restore requires the source database to be read-only during the dump, causing downtime.

Use the AWS Schema Conversion Tool (AWS SCT) to convert the schema and migrate data.

Why wrong: AWS SCT is primarily for schema conversion, not for minimizing downtime during data migration.

Use AWS DataSync to replicate the database files.

Why wrong: AWS DataSync is designed for file-based data transfers, not for live database migration with minimal downtime.

AWS Database Migration Service (AWS DMS) with the Oracle native dump and load option

Why wrong: Native dump/load does not convert schema; AWS SCT is needed for schema conversion.

AWS Server Migration Service (AWS SMS)

Why wrong: AWS SMS is for migrating on-premises servers to AWS, not for database schema conversion.

AWS Database Migration Service (AWS DMS)

Why wrong: AWS DMS migrates data but does not automatically convert schema; schema conversion must be done separately with AWS SCT.

Enable Multi-AZ on the target RDS instance.

Why wrong: Multi-AZ provides high availability, not performance improvement during migration.

Use a smaller instance class for the replication instance to reduce cost.

Why wrong: A smaller instance class may reduce performance, not improve it.

Set the LOB mode to 'Full LOB mode'.

Why wrong: Setting LOB mode to 'Full LOB mode' forces AWS DMS to load all LOB data into memory for every row, which increases per-row processing overhead and exacerbates performance issues when the source already has a high volume of write operations. This option is tempting because Full LOB mode is the correct choice when migrating tables containing large objects that must be fully materialised to avoid truncation, but here the bottleneck is write-heavy transactional load, not LOB handling.

Use AWS DataSync to transfer the MongoDB data files.

Why wrong: AWS DataSync is for file-based data, not for live database migration.

Set up a MongoDB replica set on Amazon EC2 and promote it to primary, then migrate to DocumentDB.

Why wrong: This adds complexity and does not directly migrate to DocumentDB; it's an intermediate step.

Use mongodump to export the data and mongorestore to import into DocumentDB.

Why wrong: Mongodump can impact source performance and may require downtime if consistency is needed.

Change the target database engine to Amazon RDS for MySQL.

Why wrong: Changing the target engine does not resolve the memory issue.

Disable Multi-AZ on the DMS replication instance.

Why wrong: Multi-AZ does not affect memory; it provides high availability.

Use a smaller instance class for the DMS replication instance.

Why wrong: A smaller instance class would have less memory, worsening the issue.

Deploy an Amazon EC2 instance to act as a proxy for the DMS replication.

Why wrong: DMS can connect directly to the source; a proxy is unnecessary.

Install the AWS Schema Conversion Tool on the source server to convert the schema.

Why wrong: Schema conversion is not needed since the target is also MySQL.

Add 'rds:ModifyOptionGroup' permission for the target RDS instance.

Why wrong: The `rds:ModifyOptionGroup` permission is for modifying option groups, not the maintenance window of an RDS instance.

Add 'dms:ModifyReplicationInstance' permission for the DMS replication instance.

Why wrong: The `dms:ModifyReplicationInstance` permission is for modifying the DMS replication instance itself, not the target RDS instance's maintenance window.

Add 'ec2:DescribeVpcs' permission to the policy.

Why wrong: The `ec2:DescribeVpcs` permission is not required for modifying the target RDS instance's maintenance window.

Restart the replication task with the same settings.

Why wrong: Restarting a full-load task will re-run the full load, not enable ongoing replication.

Create a new replication instance and start a new full-load task.

Why wrong: This does not enable ongoing replication; it's just a new full load.

Use AWS Schema Conversion Tool to convert the schema again.

Why wrong: Schema conversion is not related to replication type.