All Services

Analytics

Amazon Kinesis

Amazon Kinesis is AWS's real-time streaming family; because Flink and Data Firehose have separate tiles here, this guide focuses on Kinesis Data Streams as the durable ordered event-log service.

Explore pricing models, common use cases, infrastructure support, and the AWS services that commonly work with Amazon Kinesis.

Amazon Kinesis pricing and cost programs

Pricing model: Streaming ingestion and retention usage

On-Demand
Available
Reserved Instances or reserved capacity
Not applicable
Savings Plans
Not applicable
Spot
Not applicable

Billing dimensions: Stream capacity mode · Data ingested and retrieved · Retention · Enhanced fan-out

Programs and modes: On-Demand mode · Provisioned shards · Enhanced fan-out

On-Demand and provisioned stream modes use different dimensions.

Free Tier: Service-specific — verify current offers

Pricing reviewed 2026-07-25. Reviewed against the linked official AWS pricing page. Recheck regional rates and program terms before purchase.

Official AWS pricing

Official AWS sources reviewed 2026-07-21.

Why implement Amazon Kinesis?

  • Captures high-volume event records in ordered shards for multiple independent consumers, replay within retention, and near-real-time processing.
  • Offers on-demand or provisioned capacity, adjustable retention, encryption, enhanced fan-out, cross-account controls, and integrations with Lambda, Flink, Firehose, and analytics services.
  • Decouples producers from streaming consumers without operating replicated log servers or assigning every consumer its own ingestion path.

How to implement Amazon Kinesis

  1. Estimate records per second, bytes per second, partition-key distribution, consumer count, replay window, latency, and growth, then choose on-demand or provisioned mode and retention.
  2. Create an encrypted stream with least-privilege producer and consumer roles, design a versioned record envelope and high-cardinality partition key, and implement batched producers with per-record retry handling.
  3. Use Lambda or KCL consumers with checkpointing and idempotency, select shared throughput or enhanced fan-out from latency and isolation needs, and alarm on write throttling, read throttling, iterator age, errors, and shard changes.

Amazon Kinesis best practices

  • Distribute partition keys evenly and monitor hot shards, preserve original event identifiers, batch efficiently, inspect partial PutRecords failures, and use exponential backoff without reordering records that require the same key order.
  • Expect at-least-once processing, make consumers idempotent, checkpoint only durable progress, handle poison records deliberately, and ensure downstream systems can absorb replays and resharding bursts.
  • Use enhanced fan-out when consumers need dedicated per-shard throughput and lower propagation delay, monitor iterator age and quota pressure, encrypt data, use IAM roles, and test shard split or merge plus consumer failover.

Amazon Kinesis use cases and server impact

  • Clickstream, telemetry, and log ingestion
  • Real-time fraud and operational event pipelines
  • Replayable event feeds for multiple independent consumers

Replaces self-managed partitioned streaming-log infrastructure for supported workloads, while record contracts, partition-key balance, consumer idempotency, checkpoints, and downstream backpressure remain application responsibilities.

Official implementation resources

Commonly paired AWS services

Planning guides that use Amazon Kinesis