Amazon DynamoDB: The Serverless NoSQL Powerhouse

Amazon DynamoDB is a fully managed, multi-region, multi-active, durable database with built-in security, backup and restore, and in-memory caching for internet-scale applications. It is a Key-Value and Document database that delivers single-digit millisecond performance at any scale.

Core Concepts for SAA-C03

1. Data Structure: Tables, Items, and Attributes

• Items: Similar to rows in RDBMS. Each item can have different attributes (schema-less).
• Attributes: Similar to columns.
• Primary Key: Can be a simple Partition Key (PK) or a composite key (PK + Sort Key).

2. Read/Write Capacity Modes

• Provisioned Mode: You specify Reads Per Second (RCU) and Writes Per Second (WCU). Best for predictable traffic and cost control. Auto-scaling is available.
• On-Demand Mode: Scales instantly based on traffic. You pay per request. Best for unpredictable workloads or new applications.

3. Consistency Models

• Eventually Consistent Reads (Default): Maximizes read throughput; might not reflect a very recent write.
• Strongly Consistent Reads: Returns the most up-to-date data. Costs double the RCUs compared to eventual consistency.
• ACID Transactions: Support for all-or-nothing operations across multiple tables.

4. Secondary Indexes

Advanced Features & Performance

• DAX (DynamoDB Accelerator): An in-memory cache for DynamoDB. Use this when you need microseconds latency for read-heavy workloads.
• DynamoDB Streams: Captures item-level changes (Insert, Update, Delete) in real-time. Perfect for triggering Lambda functions (e.g., sending a welcome email when a user signs up).
• Global Tables: Provides multi-region, multi-active replication. Useful for disaster recovery and local performance for global users.
• TTL (Time to Live): Automatically deletes items after a specific timestamp, reducing storage costs without using WCUs.

Decision Matrix / If–Then Guide

• IF you need microseconds latency THEN use DAX.
• IF traffic is unpredictable or “spiky” THEN use On-Demand Mode.
• IF you need to search on a non-key attribute THEN use GSI (do not use Scan).
• IF you need to react to data changes in real-time THEN use DynamoDB Streams + Lambda.
• IF you need multi-region redundancy THEN use Global Tables.

Exam Tips and Gotchas

• Scan vs. Query: A Query finds items based on the Primary Key. A Scan looks at every item in the table. Always prefer Query for performance and cost.
• LSI Limitation: You cannot add an LSI to an existing table. If the exam asks how to add an index to a 3-year-old table, the answer is GSI.
• Hot Partitions: If your Partition Key is poorly designed (e.g., a “Status” field with only 2 values), one partition will get all the traffic, leading to ProvisionedThroughputExceededException.
• Large Items: DynamoDB has a 400KB item size limit. For larger files, store them in S3 and save the S3 URL in DynamoDB.

Topics covered :

Summary of key subtopics covered in this guide:

• Key-Value vs. Document models.
• Provisioned vs. On-Demand capacity modes.
• RCU/WCU calculations and consistency levels.
• LSI vs. GSI differences and limitations.
• DAX for microsecond performance.
• DynamoDB Streams for event-driven architecture.
• Global Tables for multi-region availability.
• Security (IAM, KMS, VPC Endpoints).