Streams as a cloud storage primitive

Serverless API for streaming data backed by object storage.

Preview Launch

Introducing S2

Object Storage

Object Storage

Stream Storage

Stream Storage

Blobs and byte ranges

Records and sequence numbers

PUT / GET / DELETE value of a named Object in a Bucket

APPEND / READ / TRIM records on a named Stream in a Basin

Cumbersome and expensive for granular appends

Easy and cheap to append records

Get data flowing
without the clusters

Latency flexibility

Storage classes that balance performance with cost.

Standard< 500 msExpress< 50 ms

Always durable on object storage in multiple zones.

Unlimited streams

Model your domain naturally –
say goodbye to partition limits and fixed costs.

Elastic throughput

Reactive scaling to GiBps – and down to zero.
Each stream can push up to 125 MiBps writes.

Versatile building block
for data system patterns

Elastic throughput and bottomless storage mean producers can burst and consumers can read at their desired pace. Pay for actual usage, not peak capacity.

Activity like impressions and clicks

Application and security events

Metric data points

Frequently Asked Questions