DynamoDB Streams 与全局表

Sometimes you need to know the moment data changes — to update a cache, send a notification, or sync another system.

DynamoDB Streams capture an ordered log of every item-level change in a table.

Each stream record describes an INSERT, MODIFY, or REMOVE. You choose what data the record carries via the stream view type:

• KEYS_ONLY
• NEW_IMAGE
• OLD_IMAGE
• NEW_AND_OLD_IMAGES

Enable a stream on a table and pick a view type.

The most common pattern is connecting a stream to a Lambda function. Lambda is invoked in batches as records arrive, letting you process changes serverlessly.

Streams power many patterns:

• Keeping a search index (OpenSearch) in sync
• Aggregating analytics
• Sending notifications on changes
• Replicating to another data store

Streams are organized into shards. Records within a shard are strictly ordered by the sequence in which changes occurred, which preserves per-item change order.

Global Tables replicate your DynamoDB table across multiple AWS regions automatically, giving users low-latency access no matter where they are.

Global Tables are active-active: you can read and write in any region, and changes replicate to the others, typically within a second.

Global Tables are built on top of DynamoDB Streams under the hood.

When the same item is written in two regions at once, DynamoDB resolves the conflict using a last writer wins strategy based on timestamps.

Reach for Global Tables when you need:

• Low latency for a worldwide user base
• Regional disaster recovery
• Multi-region active-active applications

Remember the distinction:

• Streams — react to changes within your own systems
• Global Tables — replicate the table itself across regions

Global Tables use Streams internally to do their work.

You learned advanced DynamoDB features:

• Streams capture ordered item-level changes
• Streams commonly trigger Lambda
• Global Tables replicate active-active across regions
• Conflicts resolve with last writer wins

Together they enable real-time, globally distributed applications.