Question: Is DynamoDB serverless?

Answer

Yes. Amazon DynamoDB is a serverless database: there are no servers, instances or clusters to provision, patch or resize. You create a table and call its HTTPS API, and AWS runs the storage, replication across multiple Availability Zones and scaling behind it. AWS itself describes DynamoDB as a serverless, fully managed NoSQL database.

How serverless your table is depends on its capacity mode.

• On-demand, the default and the mode AWS recommends for most workloads since the November 2024 price cut, bills per read and write request plus storage. Nothing is provisioned, the table absorbs traffic changes on its own, and an idle table costs only its storage (plus backups or streams if you turn them on). You can cap spend with a maximum throughput setting.
• Provisioned mode bills by the hour for the read and write capacity units you choose, used or not, optionally adjusted by auto scaling. There are still no servers, but you are planning capacity again.

What serverless doesn't mean.

• No limits. Each partition serves at most 3,000 read units and 1,000 write units per second, and each table has adjustable default quotas of 40,000 read and 40,000 write units per second. A single hot key can throttle even an on-demand table; see DynamoDB hot partitions and DynamoDB limits.
• No bill at zero. Storage beyond the 25 GB free tier is charged per GB-month whether or not anyone reads it.
• Connection management. DynamoDB has no connection pools or cold starts to manage: every request is a stateless HTTPS call. That is why it pairs naturally with AWS Lambda and other short-lived compute.

DynamoDB capacity modes compares the two billing models in detail, and the DynamoDB pricing calculator estimates what each would cost for your traffic.