What’s New in Amazon DynamoDB (2025): Features, Roadmap, and Multi-Attribute Keys

DynamoDB Innovation: Evolution and Current State
Speaker: Mark Pinaud, Senior Manager, Product Management, AWS DynamoDB

In this DynamoDB Day session, AWS dives into the latest DynamoDB launches, performance improvements, global table updates, generative-AI tooling, and a first look at upcoming enhancements like multi-attribute partition and sort keys for GSIs.

If you’re building high-scale workloads on DynamoDB—or modernizing existing ones—this session gives you a clear view of where DynamoDB is today and where it's heading next.

Chapters
0:00 – Welcome
0:03 – Speaker intro & background
0:32 – What hasn’t changed in DynamoDB
1:39 – DynamoDB scale benchmarks
2:14 – Why over 1M customers use DynamoDB
2:27 – Innovation history & major launches
3:06 – How AWS prioritizes DynamoDB feature work
3:44 – What DynamoDB invests in next
4:15 – Recent launches overview
4:28 – Warm Throughput
4:46 – 50% on-demand price reduction
5:08 – Global Tables price reduction (67%)
5:30 – Strong consistency now GA
5:42 – Configurable PITR windows
6:16 – Latency percentile stats
6:44 – Granular throttling exceptions
7:10 – Throttles-only Contributor Insights
7:59 – Security updates
8:33 – Zero-ETL integrations overview
8:56 – Generative AI: DynamoDB MCP Server
9:15 – Data modeling + validation MCP tools
10:06 – Database Analyzer MCP
10:35 – LangGraph connector
11:01 – More AWS gen-AI integrations coming
11:27 – NEW upcoming feature: Multi-attribute keys
11:40 – Why they matter for hierarchical queries
12:30 – How GSIs work today vs. new model
13:36 – Querying examples with new partition/sort keys
14:52 – Cost & storage benefits
16:02 – Expected rollout
16:08 – DynamoDB @ re:Invent 2025 sessions
17:10 – Live Q&A begins
17:44 – Anything Mark wanted to add?
18:24 – Follow-up with Alex DeBrie
19:17 – Q&A: Strongly consistent GSI reads
20:32 – Q&A: Cost impact of multi-attribute keys
21:51 – Q&A: DynamoDB MCP for config & cost management
24:10 – Q&A: When to use provisioned vs. on-demand
26:25 – Q&A: On-demand scaling considerations
28:02 – Warm Throughput for event spikes
29:36 – Q&A: Migrating to multi-attribute keys
41:03 – Q&A: When to pair DynamoDB with another DB
43:04 – Q&A: Using PartiQL
44:41 – Closing insights