Table Classes | DynamoDB cost optimization

What are DynamoDB table classes?

DynamoDB offers two table classes designed to help you optimize for cost: Standard and Standard-Infrequent Access (Standard-IA).¹ Each table class offers different pricing for data storage, read requests, and write requests. The table class you choose applies to the entire table and all its secondary indexes—you cannot mix table classes within a single table.

Table class comparison

Feature	Standard	Standard-IA
Storage cost	Base rate	~60% lower than Standard
Read/Write cost	Base rate	~25% higher than Standard
Best for	Throughput-dominant workloads	Storage-dominant workloads
Reserved capacity	Supported	Not supported

Key principle: Standard-IA trades higher per-request costs for dramatically lower storage costs.²

When to use each table class

Use Standard table class when

Storage represents less than 50% of your total monthly DynamoDB cost. This indicates a workload that regularly accesses or updates stored data.³

Typical use cases:

Active user sessions or shopping carts
Real-time analytics or dashboards
High-velocity transaction tables
Gaming leaderboards and active player data
IoT telemetry for current devices
Tables with frequent reads and writes

Use Standard-IA table class when

Storage represents more than 50% of your total monthly DynamoDB cost. This indicates data that is created and retained but rarely accessed.³

Typical use cases:

Application logs and audit trails
Order history and transaction archives
Old social media posts or comments
Past gaming achievements and historical scores
Compliance data with long retention requirements
Infrequently accessed reference data
Customer interaction history

Cost savings with Standard-IA

Storage cost reduction

Standard-IA delivers ~60% lower storage costs compared to Standard, making it ideal for tables storing large volumes of rarely-accessed data.²

Example: 1TB of infrequently accessed data

Using approximate us-east-1 pricing:

Table Class	Storage Cost/GB/Month	1TB Monthly Storage Cost	Annual Storage Cost
Standard	$0.25	$250	$3,000
Standard-IA	$0.10	$100	$1,200
Savings	-	$150/month	$1,800/year

Throughput cost impact

Standard-IA charges ~25% more per read/write request than Standard.³ The break-even point depends on your storage-to-throughput ratio.

Switching between table classes

Key facts about switching

No downtime: Your table remains fully available during the switch¹
No code changes: Your application continues to work without modifications¹
Background process: The switch happens asynchronously; time depends on table size and traffic⁴
Change limit: Maximum 2 table class changes per 30-day period⁴
Applies to indexes: All GSIs and LSIs inherit the new table class¹

Limitations and considerations

Reserved capacity

Standard-IA tables do not support reserved capacity.³ If you currently use reserved capacity with Standard tables, switching to Standard-IA means you’ll lose that discount.

Recommendation: If you have significant reserved capacity commitments, stay on Standard until the reservation expires, then reevaluate.

Global tables

Both Standard and Standard-IA support global tables, but remember:

All replicas must use the same table class¹
Changing table class affects all regions simultaneously
Replication costs (rWCUs) scale with your throughput, not your table class

Table class changes take time

The table class update is a background process that doesn’t block table operations.⁴ However:

Large tables (hundreds of GB or more) may take hours to complete
You can monitor progress in the console or via DescribeTable API
No more than 2 changes allowed per 30-day period⁴

Design strategies for cost optimization

Strategy 1: Separate hot and cold data

Instead of mixing frequently and infrequently accessed data in one table, split them into separate tables with different table classes.

Example: E-commerce order system

Orders table (Standard):

Recent orders (last 90 days)
High read/write activity
Supports real-time order tracking

OrderHistory table (Standard-IA):

Orders older than 90 days
Rare reads (customer account history, tax reporting)
Lower storage costs for millions of historical orders

Strategy 2: Use TTL before archiving

Combine Time to Live (TTL) with Standard-IA to automatically clean up expired data while keeping important data cheap.

Strategy 3: Start with Standard, optimize later

For new tables, start with Standard until you have real cost data:

Launch: Create table with Standard class
Monitor: Track costs for 1-3 months using Cost Explorer
Analyze: Calculate storage vs throughput ratio
Optimize: Switch to Standard-IA if storage > 50% of costs

This approach avoids premature optimization and lets actual usage patterns guide your decision.

Does Standard-IA affect performance?

No. Standard-IA offers the same performance, durability, and availability as Standard tables.⁴ Both classes:

Support the same read/write throughput
Provide single-digit millisecond response times
Offer 99.99% availability (99.999% for global tables)
Use the same partition architecture