Capacity and Pricing

Understanding DynamoDB's capacity modes and pricing model is essential for controlling costs and ensuring performance. This lesson covers on-demand and provisioned capacity, auto-scaling, reserved capacity, and how to estimate costs.

---

Read and Write Capacity Units

DynamoDB measures throughput in capacity units:

Read Capacity Units (RCUs)

Read Type	Size	RCUs
Strongly consistent	Up to 4 KB	1 RCU
Eventually consistent	Up to 4 KB	0.5 RCU
Transactional	Up to 4 KB	2 RCUs

For items larger than 4 KB, DynamoDB rounds up. For example, a 6 KB item consumes 2 RCUs for a strongly consistent read.

Write Capacity Units (WCUs)

Write Type	Size	WCUs
Standard	Up to 1 KB	1 WCU
Transactional	Up to 1 KB	2 WCUs

For items larger than 1 KB, DynamoDB rounds up. For example, a 3 KB item consumes 3 WCUs.

---

Capacity Modes

DynamoDB offers two capacity modes:

On-Demand Mode

Pay per request with no capacity planning:

Feature	Description
Billing	Per read/write request
Scaling	Instant — handles spikes automatically
Provisioning	None required
Best for	Unpredictable workloads, new applications, spiky traffic

aws dynamodb create-table \
  --table-name MyTable \
  --billing-mode PAY_PER_REQUEST \
  ...

Provisioned Mode

You specify the number of reads and writes per second:

Feature	Description
Billing	Per provisioned RCU/WCU per hour
Scaling	Manual or auto-scaling
Provisioning	Set RCU and WCU values
Best for	Predictable workloads, cost optimisation

aws dynamodb create-table \
  --table-name MyTable \
  --billing-mode PROVISIONED \
  --provisioned-throughput \
    ReadCapacityUnits=100,WriteCapacityUnits=50 \
  ...

Mode Comparison

Aspect	On-Demand	Provisioned
Capacity planning	None	Required
Cost per request	Higher	Lower
Scaling speed	Instant	Minutes (auto-scaling)
Burst handling	Automatic	Limited by provisioned capacity
Billing model	Pay per request	Pay per hour
Switching	Can switch once every 24 hours	Can switch once every 24 hours

---

Auto-Scaling

Auto-scaling automatically adjusts provisioned capacity based on actual traffic:

How It Works

1. You set a target utilisation (e.g., 70%)
2. You set minimum and maximum capacity
3. DynamoDB uses Application Auto Scaling to adjust capacity
4. When utilisation exceeds the target, capacity scales up
5. When utilisation drops, capacity scales down

Configuration (AWS CLI)

# Register the scalable target
aws application-autoscaling register-scalable-target \
  --service-namespace dynamodb \
  --resource-id "table/MyTable" \
  --scalable-dimension "dynamodb:table:ReadCapacityUnits" \
  --min-capacity 5 \
  --max-capacity 1000

# Create the scaling policy
aws application-autoscaling put-scaling-policy \
  --service-namespace dynamodb \
  --resource-id "table/MyTable" \
  --scalable-dimension "dynamodb:table:ReadCapacityUnits" \
  --policy-name "ReadAutoScaling" \
  --policy-type "TargetTrackingScaling" \
  --target-tracking-scaling-policy-configuration '{
    "TargetValue": 70.0,
    "PredefinedMetricSpecification": {
      "PredefinedMetricType": "DynamoDBReadCapacityUtilization"
    }
  }'

Tip: Auto-scaling reacts to sustained changes, not short spikes. It typically takes 1-2 minutes to scale up. For sudden bursts, DynamoDB provides burst capacity from unused capacity (up to 5 minutes of unused throughput).

---