Amazon DynamoDB: The Serverless NoSQL Workhorse

TL;DR

Amazon DynamoDB is AWS’s fully managed NoSQL database — infinite scale, single-digit millisecond latency, and true serverless pricing. It’s the default choice for key-value and document workloads on AWS. The on-demand mode means you pay per request, not for idle capacity. The catch: query patterns must be designed upfront (no ad-hoc queries), and costs can spiral with hot partitions or large scans. For apps that fit its access patterns, it’s unbeatable. For complex relational queries, look elsewhere.

What Is It?

DynamoDB is a fully managed NoSQL database service that provides fast and predictable performance with seamless scalability.

Core Concepts

┌─────────────────────────────────────────────────────────────┐
│                    DynamoDB Table                            │
│                                                              │
│   Partition Key (UserID) │ Sort Key (Timestamp) │ Attributes │
│   ───────────────────────┼─────────────────────┼────────────│
│   user123                │ 2024-01-15T10:00:00 │ name, email │
│   user123                │ 2024-01-15T11:00:00 │ action      │
│   user456                │ 2024-01-15T10:00:00 │ name, email │
│                                                              │
│   Primary Key = Partition Key + Sort Key (composite)        │
└─────────────────────────────────────────────────────────────┘

Key Features

Feature Description
Tables Collection of items (like a table, but schema-less)
Items Individual records (max 400 KB)
Attributes Named data elements
Primary Key Partition key alone, or partition + sort key
Secondary Indexes Global (any attributes) or Local (same partition key)

Capacity Modes

Mode Use Case Pricing
On-Demand Unknown/unpredictable traffic Per request
Provisioned Predictable traffic Per hour + auto-scaling

Pricing

On-Demand Pricing

Operation Price (per million)
Write request unit (WRU) $1.25
Read request unit (RRU) $0.25
Transaction write $2.50
Transaction read $0.50

1 WRU = 1 KB write 1 RRU = 4 KB eventually consistent read (or 0.5 KB strongly consistent)

Storage

Tier Price/GB/Month
Standard $0.25
Infrequent Access $0.10

Cost Example: 10M writes, 100M reads, 100 GB storage

Component Cost
Writes (10M) $12.50
Reads (100M) $25.00
Storage (100 GB) $25.00
Total $62.50

GCP Alternative: Firestore

Feature DynamoDB Firestore Winner
Model Key-value + document Document Tie
Queries Limited Richer Firestore
Transactions Yes Yes Tie
Realtime DynamoDB Streams Native Firestore
Pricing Per request Per operation Similar
Offline sync No Yes (mobile) Firestore

Firestore Data Model

Collection: users
├── Document: user123
│   ├── name: "John"
│   └── orders (subcollection)
│       ├── order456
│       └── order789

Azure Alternative: Cosmos DB

Feature DynamoDB Cosmos DB
APIs DynamoDB MongoDB, Cassandra, SQL, Gremlin
Latency Single-digit ms Single-digit ms
Multi-region Global Tables Built-in
Pricing Simpler Complex

Real-World Use Cases

Use Case 1: Session Store

Challenge: 10M active sessions, sub-millisecond access

Architecture:

User Request → API Gateway → Lambda
                    ↓
            DynamoDB (Session table)
                Partition: SessionID
                TTL: 24 hours

Use Case 2: Shopping Cart

Challenge: High write throughput, consistent reads

Table: Carts
├── Partition: UserID
├── Sort: ItemID
└── Attributes: quantity, price, timestamp

Use Case 3: Time-Series Data

Challenge: IoT sensor data, high ingest

Table: SensorData
├── Partition: DeviceID
├── Sort: Timestamp
└── Attributes: temperature, humidity

The Catch

1. Query Limitations

2. Hot Partitions

One partition key with heavy traffic:

3. Cost Surprises

4. 400 KB Item Limit

Large objects require:

Verdict

Grade: A

Best for:

When to use:

When not to use:

Researcher 🔬 — Staff Software Architect