Module 6 cheat sheet
One-page revision summary for Databases & Data Architecture: services, Azure mapping, CLI commands, limits, decision rules, exam traps and cost traps.
Assumes: AWS CLI v2 · AWS SDK for .NET v4 · EF Core 10 · Npgsql 10 · PostgreSQL 17 · Region us-east-1 · facts as stated in the Module 6 lessons; items marked VERIFY there are not repeated as certain here. Verified against official sources on 2026-11-18.
A printable summary of Module 6. Each row links back to the lesson with the detail.
Key services
| Service | What it is for |
|---|---|
| RDS for SQL Server / PostgreSQL | Managed relational engine; you own network, identity, keys, parameters and schema |
| Aurora (PostgreSQL, MySQL) | Shared cluster volume, up to 15 readers, Serverless v2, Global Database |
| Babelfish for Aurora PostgreSQL | T-SQL and TDS compatibility layer; subset only, assess first |
| DynamoDB | Key-value at scale for known access patterns: idempotency records, lookups, audit |
| ElastiCache (Valkey / Redis OSS) | Sub-millisecond cache, counters, sessions; never a system of record by default |
| DocumentDB | MongoDB-compatible (not MongoDB) documents: catalogues, flexible forms |
| Redshift / S3 + Athena | Warehouse and lake for history and analytics, off the OLTP database |
| RDS Proxy | Connection multiplexing, faster failover handling; paid per vCPU-hour |
| DMS + DMS Schema Conversion | Full load + CDC, and T-SQL to PostgreSQL conversion with assessment |
| KMS / Secrets Manager | Customer managed key at rest; rotated credentials |
| CloudWatch Database Insights | DB load by wait event and top SQL (replaced the Performance Insights console) |
.NET / Azure → AWS
| Microsoft / Azure | AWS |
|---|---|
| SQL Server / Azure SQL MI | RDS for SQL Server |
| Azure SQL Database serverless | Aurora PostgreSQL Serverless v2 |
| Always On AG (sync secondary) | Multi-AZ DB instance (standby not readable) |
| Always On readable secondaries | Multi-AZ DB cluster or read replicas |
sp_configure, SQL Agent options | Parameter groups and option groups |
| Cosmos DB | DynamoDB |
| Azure Cache for Redis | ElastiCache for Valkey |
| Synapse dedicated pool | Redshift / Redshift Serverless |
| Managed identity to Azure SQL | IAM database authentication (not SQL Server) |
| Key Vault TDE key | KMS customer managed key, chosen at creation |
| DMA / SSMA | DMS Schema Conversion + DMS |
| Query Store / Query Performance Insight | Database Insights + pg_stat_statements / Query Store |
Most useful CLI commands
| Task | Command |
|---|---|
| Engine versions available | aws rds describe-db-engine-versions --engine postgres |
| Instance status and settings | aws rds describe-db-instances --db-instance-identifier ID |
| Wait until ready | aws rds wait db-instance-available --db-instance-identifier ID |
| Manual snapshot | aws rds create-db-snapshot --db-instance-identifier ID --db-snapshot-identifier SNAP |
| Point-in-time restore (new instance) | aws rds restore-db-instance-to-point-in-time ... |
| Restore from snapshot | aws rds restore-db-instance-from-db-snapshot ... |
| Aurora cluster restore | aws rds restore-db-cluster-to-point-in-time ... |
| Inspect proxies | aws rds describe-db-proxies |
| Create DynamoDB table | aws dynamodb create-table ... |
| Query DynamoDB | aws dynamodb query --table-name T --key-condition-expression ... |
| Enable TTL | aws dynamodb update-time-to-live --table-name T --time-to-live-specification ... |
| Check a secret | aws secretsmanager describe-secret --secret-id ID |
| Clean up | aws rds delete-db-instance --db-instance-identifier ID --skip-final-snapshot (lab only), then delete snapshots and parameter groups |
Key limits and defaults
| Item | Value | Lesson |
|---|---|---|
| Automated backup retention | 0 to 35 days; 0 disables PITR; console default 7, CLI default 1 | Multi-AZ, replicas, backups, PITR |
| PITR | Any second in the window, typically within about 5 minutes of now, always to a new instance | Multi-AZ, replicas, backups, PITR |
| Manual snapshot quota | 100 per Region | Multi-AZ, replicas, backups, PITR |
| Multi-AZ instance failover | Typically 60 to 120 seconds | Multi-AZ, replicas, backups, PITR |
| RDS read replicas | Up to 15 per primary; SQL Server Enterprise only | Multi-AZ, replicas, backups, PITR |
| Aurora storage and readers | Six copies across three AZs; writer plus up to 15 readers | Amazon Aurora |
| Aurora Global Database | One writer Region, up to 10 secondaries, typically under 1 s lag | Amazon Aurora |
| gp3 baseline | 3,000 IOPS and 125 MiB/s (PostgreSQL rises to 12,000 and 500 at 400 GiB) | RDS |
| IAM database auth token | Valid 15 minutes; PostgreSQL, MySQL, MariaDB only | Encryption and database access |
| RDS Proxy timeouts | Idle client 1,800 s; maximum connection life 24 h | Connection pooling and RDS Proxy |
| DynamoDB item size | 400 KB including attribute names | DynamoDB fundamentals |
| DynamoDB Query/Scan page | Up to 1 MB per call; paginate with LastEvaluatedKey | DynamoDB fundamentals |
TransactWriteItems | Up to 100 actions, 4 MB, one Region, each item once, double capacity cost | DynamoDB for banking correctness |
| DynamoDB Streams retention | 24 hours | DynamoDB for banking correctness |
| PostgreSQL error codes | 40001 serialization failure: retry the whole transaction; 25006 read-only (failover) | Transactions · EF Core on AWS |
Decision rules
| If… | Then… |
|---|---|
| Joins, integrity, ACID ledger | Relational: RDS or Aurora PostgreSQL |
| T-SQL heavy, licences fixed | RDS for SQL Server (licence cost drives the decision) |
| T-SQL heavy in-house app, wants lower licence cost | Assess Babelfish, then test; do not assume |
| Known key lookups, idempotency, huge scale | DynamoDB with on-demand capacity |
| Hot reads, rate limits, sessions | ElastiCache for Valkey with HybridCache; fail-open or fail-closed decided per use |
| Flexible JSON, no money | DocumentDB, or jsonb on an existing Aurora cluster |
| Years of history, ad hoc analysis | S3 + Athena; Redshift for high-concurrency complex SQL |
| Lambda or bursty autoscaling exhausts connections | Budget connections, shrink pools, then add RDS Proxy |
| No password wanted (PostgreSQL, MySQL, MariaDB) | IAM database authentication; otherwise Secrets Manager with alternating-users rotation |
| Lost updates | Guarded atomic UPDATE ... WHERE balance >= @amount, or a concurrency token |
| Duplicate payment risk | Unique key plus stored response, claimed in the same transaction as the money movement |
| Migrations in CI/CD | Bundle or script, one-shot job, direct connection (not the proxy) |
| Migration cut-over | DMS full load + CDC, independent validation, agreed rollback |
Common exam traps
- Multi-AZ is not a backup. A bad delete replicates; use PITR or snapshots.
- The Multi-AZ instance standby does not serve reads; read replicas and Aurora readers do.
- Replicas are asynchronous: a read just after a write may be stale.
- Encryption at rest is chosen at creation. Encrypt an unencrypted instance by restoring a snapshot copy made with a key.
- Snapshots under the
aws/rdsmanaged key cannot be shared; use a customer managed key. - IAM database authentication does not exist for RDS for SQL Server.
- Strongly consistent DynamoDB reads are never available on a GSI; GSIs are eventually consistent.
- TTL deletes eventually, not on time: filter expired items on read.
- A global table or transaction is not a cross-Region transaction: transactions stay in one Region.
- PITR restores to a new instance, never over the existing one.
- Disabling the KMS key stops the database.
- DocumentDB is MongoDB-compatible, not MongoDB.
- Performance Insights console is replaced by CloudWatch Database Insights; the PI API continues.
Cost traps
| Trap | Avoid it by |
|---|---|
| RDS for SQL Server licence-included | Design-only; use Developer edition in Docker |
| RDS, Aurora and proxies left running | Deploy and destroy the same day; stop is temporary |
| Manual snapshots and old restored instances | Delete after drills; they bill for storage |
| RDS Proxy billed per vCPU-hour of the target | Add it only where connection exhaustion is measured |
| Aurora Global Database, DocumentDB, Redshift | Design-only exercise with diagrams and an ADR |
| DMS replication instance running after rehearsal | Stop and delete tasks, endpoints and the instance |
| DynamoDB storage growth and scans | TTL, archive to S3, query by key |
| Athena scanning raw CSV | Parquet, partitions, workgroup limits |
| ElastiCache idle cluster | Local Valkey in Docker for learning |
| Secrets Manager secrets forgotten | Delete with the lab; use Parameter Store for throwaway labs |
Independent learning resource — not affiliated with, endorsed or sponsored by Amazon Web Services. AWS and related marks are trademarks of Amazon.com, Inc. or its affiliates. All banking data in examples is fictional. Sample architectures are for learning and are not automatically compliant with GDPR, PCI DSS or any banking regulation.
Lab 6: Harbourline banking data store (RDS PostgreSQL, EF Core, IAM auth, restore drill)
Build and run the module project: an idempotent, concurrency-safe ledger API on EF Core and PostgreSQL, first locally, then on an encrypted private RDS instance with IAM authentication, ending in a measured, evidenced point-in-time restore drill and a written decision record.
Module 7: Data Engineering & ETL Pipelines
Design and build secure batch and streaming data pipelines on AWS, from an S3 data lake through Glue, Athena and Redshift to reporting. The module ends with a full banking transaction pipeline that validates, masks, reconciles and audits fictional transaction data, plus a near-real-time fraud-signal variant.