AWS DBS-C01 Study Guide

The AWS Certified Database - Specialty (DBS-C01) is an advanced certification that rewards deep technical knowledge of AWS database services. Unlike associate-level exams, the Database Specialty expects you to make sophisticated architecture decisions, understand service-specific trade-offs, and solve complex scenarios involving high availability, disaster recovery, and performance optimization.

This guide focuses on the core concepts, services, and patterns you need to pass. It's designed for professionals with production experience who want to solidify their knowledge before the exam.

Exam Domains

Management and Operations is the heaviest weight, followed closely by design. The exam expects you to know operational details: backups, failover, parameter groups, monitoring, and how to troubleshoot in production.

Key Services to Know

Amazon RDS

RDS supports five database engines: MySQL, PostgreSQL, Oracle, SQL Server, and MariaDB. The exam tests your ability to choose between them and configure them correctly.

• Engine trade-offs: PostgreSQL is open-source and feature-rich. MySQL is simpler and faster for some workloads. Oracle and SQL Server command premium licensing. MariaDB is MySQL-compatible and open.
• Multi-AZ deployments: Automatic failover across availability zones. Know that failover takes 1-2 minutes and incurs a brief connection loss.
• Read replicas: Scale read workloads. Can be in the same AZ, different AZ, or different region. Replication lag is possible; know how to monitor it.
• Backups and snapshots: Automated backups are retained for 1-35 days (configurable). Manual snapshots are kept indefinitely. Backup window and maintenance window are separate.
• Enhanced Monitoring and Performance Insights: Essential for troubleshooting performance issues. Enhanced Monitoring gives OS-level metrics; Performance Insights shows active sessions and wait events.
• Parameter groups and option groups: Parameter groups control database configuration. Option groups enable database features like native backup/restore on Oracle.

Amazon Aurora

Aurora is AWS-native and decouples storage from compute. It's increasingly the right answer for "high-availability relational database."

• Aurora vs RDS MySQL/PostgreSQL: Aurora offers faster failover (seconds vs minutes), automatic storage scaling, and read replicas without replication lag.
• Aurora Global Database: Active-active replication across regions for disaster recovery. RPO of seconds, RTO of seconds.
• Reader and custom endpoints: Route read-heavy queries to read replicas. Custom endpoints let you isolate workloads (analytics queries on one set of replicas, OLTP on another).
• Aurora Serverless: Auto-scaling compute for unpredictable workloads. You pay only for what you use, but query latency is higher due to compute scaling overhead.
• Backtrack: Time-based recovery without restores. You can rewind a database to any point in the last 72 hours (configurable), then resume operations. This is unique to Aurora.
• Parameter groups: Aurora uses cluster parameter groups (shared across all instances in the cluster) and instance parameter groups (for per-instance customization).

Amazon DynamoDB

DynamoDB is fully managed NoSQL. The exam tests when it's the right choice and how to configure it for performance and cost.

• Partition and sort keys: Partition key determines which partition the item goes to. Sort key orders items within a partition. Choose these carefully for your access patterns.
• On-demand vs provisioned capacity: On-demand scales automatically but costs more per request. Provisioned is cheaper if you can predict traffic. Burst capacity allows short spikes above provisioned.
• Indexes: Global Secondary Indexes (GSI) have their own partition and sort key, independent of the base table. Local Secondary Indexes (LSI) share the partition key with the base table and are limited to 10 GB per partition key value.
• DynamoDB Streams: Capture changes (new item, modified item, removed item) as stream records. Use with Lambda for event-driven workflows.
• DynamoDB Accelerator (DAX): In-memory cache for microsecond latency. Useful for read-heavy workloads.
• Capacity calculation: One read capacity unit (RCU) = 4 KB of strongly consistent reads per second, or 8 KB of eventually consistent reads per second. One write capacity unit (WCU) = 1 KB per second. The exam includes RCU/WCU calculations.
• Point-in-time recovery (PITR): Restore to any point in the last 35 days. On-demand only.
• Time-to-live (TTL): Automatically delete items after expiration. Useful for session data or temporary records.

Amazon Redshift

Redshift is a data warehouse. It's columnar, compressed, and optimized for analytical queries on massive datasets.

• Cluster architecture: A cluster consists of a leader node and compute nodes. The leader parses and optimizes queries; compute nodes execute them in parallel.
• Node types: Dense compute (RA3) nodes have managed storage that scales independently. Dense storage (ds2) nodes have fixed SSD storage.
• Distribution styles: EVEN distributes rows round-robin. KEY distributes by a column value (for joins). ALL replicates the table to every node (for small dimension tables).
• Sort keys: Define the physical order of rows on disk. Use for range queries and time-series data. Compound sort keys use multiple columns; interleaved sort keys optimize for different columns.
• Redshift Spectrum: Query data directly in S3 without loading. Useful for cold historical data.
• Concurrency scaling: Add compute capacity automatically for query spikes. You pay per second of scaling.
• Enhanced VPC routing: Route traffic through your VPC rather than AWS's public network. Required for compliance in some environments.
• Snapshots and disaster recovery: Automated snapshots are retained for 1-35 days. Cross-region snapshots are available for disaster recovery.

Amazon ElastiCache

ElastiCache provides in-memory caching with two engines: Memcached and Redis.

• Memcached: Simple key-value store. Good for session storage and caching. No persistence. Auto-discovery lets applications discover cache nodes dynamically.
• Redis: Supports strings, lists, sets, hashes, and streams. Persistence (snapshots and AOF). Replication for high availability. Pub-sub messaging.
• Cluster mode enabled vs disabled: Cluster mode disabled is a primary with replicas; all data is on every replica. Cluster mode enabled partitions data across shards, reducing per-node memory requirements and increasing throughput.
• Multi-AZ and automatic failover: Both engines support Multi-AZ with automatic failover. Failover takes seconds.
• Connection pooling: ElastiCache has a maximum number of connections. Use connection pooling in your application to avoid exhausting the limit.

Other AWS Database Services

• Amazon DocumentDB: MongoDB-compatible NoSQL database. Fully managed, with ACID transactions, backup/restore, and encryption.
• Amazon Neptune: Fully managed graph database. Use for social networks, recommendation engines, and knowledge graphs.
• AWS Database Migration Service (DMS): Heterogeneous database migration. Supports full load and change data capture (CDC). Can run locally or on EC2/RDS.
• Amazon Keyspaces: Apache Cassandra-compatible wide-column database. Fully managed with on-demand and provisioned billing.
• Amazon Timestream: Time-series database optimized for metrics. Automatic data aggregation and expiration.

Architecture Patterns

High Availability

Spread databases across at least 2 availability zones:

• Multi-AZ RDS or Aurora for automatic failover.
• Read replicas in different AZs to distribute read load.
• ElastiCache with Multi-AZ and automatic failover.
• DynamoDB is multi-AZ by default.

Disaster Recovery

For critical workloads:

• Aurora Global Database for active-active replication across regions.
• Redshift cross-region snapshots for point-in-time recovery.
• RDS cross-region read replicas (read-only; promote for failover).
• DMS for heterogeneous replication (e.g., on-premises to AWS).

Performance Optimization

• Use read replicas for read-heavy workloads.
• DynamoDB GSI for alternative query patterns without scanning the base table.
• ElastiCache or DAX for hot data that's read frequently.
• Redshift for analytical queries on large historical datasets.

Cost Optimization

• On-demand DynamoDB for unpredictable traffic; provisioned for predictable traffic.
• Aurora Serverless for workloads with extended idle periods.
• Redshift for massive analytical workloads (cheaper than running transactional databases).
• Reserved capacity for RDS and Redshift if you can commit.

Common Exam Traps

Confusing Multi-AZ and read replicas: Multi-AZ is synchronous failover. Read replicas are asynchronous and can be in different regions. Questions ask "which solves HA" vs "which scales reads." Know the difference.

DynamoDB index limits: LSI has a 10 GB per partition key value limit. GSI has no limit. Picking LSI when you should use GSI is wrong.

RDS automated backup retention: Backups are retained for 1-35 days (default 7). You can't restore to before the retention window. Snapshots are manual and kept indefinitely.

Redshift vs RDS: Redshift is a data warehouse for analytical queries. RDS is for transactional, relational workloads. Questions exploit this distinction.

Aurora endpoints: Aurora has writer endpoints (for the primary instance), reader endpoints (for replicas), and custom endpoints (for groups of replicas). Know which to use when.

DynamoDB eventual consistency: Eventually consistent reads are half the cost of strongly consistent reads. But "real-time" scenarios often require strong consistency. Know the trade-off.

Encryption inheritance: Some services inherit encryption from parent resources. For example, RDS automated backups inherit encryption from the DB instance. Know which services do and don't.

Study Plan (6 Weeks)

Each week should include 2-3 hours of study daily. Allocate time for hands-on lab work: deploy RDS, Aurora, and DynamoDB instances in your own AWS account and test failover, scaling, and backup behavior.

Practice Exam Strategy

• Read all answers: Elimination matters. AWS test writers craft distractor answers that feel plausible.
• Look for keywords: "Most cost-effective," "minimal latency," "automatic failover," "zero downtime." Keywords reveal the question's intent.
• Scenario-based context: The exam presents architecture scenarios and asks you to choose the right service or configuration. Focus on the constraints: time-to-recover, data consistency, cost, throughput.
• Understand why answers are wrong: When you miss a question, spend time understanding not just why the correct answer is right, but why the other three are wrong. This reveals the nuances tested by AWS.
• Time management: You have 170 minutes for 65 questions. That's 2.6 minutes per question. Use 1.5-2 minutes per question, flag difficult ones, and return to them if time allows.

Use our DBS-C01 practice exams to build speed and confidence. Aim for 80%+ before booking your real exam.

Final Thoughts

The Database Specialty is rewarding for professionals who design and operate databases on AWS. Unlike associate exams, it's not about memorizing facts. It's about understanding trade-offs: RDS vs DynamoDB, Multi-AZ vs read replicas, provisioned vs on-demand, encryption at rest vs in transit.

Study production scenarios. Think about how you'd design a high-availability e-commerce database, a real-time analytics platform, a multi-tenant SaaS application. The exam tests your ability to make these decisions under constraints.

Master the core services: RDS, Aurora, DynamoDB, and Redshift. Understand their strengths, limitations, and use cases. Know encryption, backup, and replication inside out. Practice until scenario-based questions feel natural.

Good luck on the exam. You've got this.