All Services

Analytics

Amazon Data Firehose

Amazon Data Firehose buffers, optionally transforms or converts, and delivers streaming records to S3, Redshift, OpenSearch, Apache Iceberg tables, Splunk, and supported HTTP or partner destinations.

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

Streaming delivery cost and buffering design

Amazon Data Firehose Pricing & Delivery Guide

Firehose is delivery, not storage: it buffers records by size or time and writes them to a destination. Ingested volume is the base charge, and format conversion, VPC delivery, and dynamic partitioning each add their own dimension.

Official AWS sources reviewed 2026-08-29.

Is Kinesis Data Firehose now Amazon Data Firehose?

AWS documents the service as Amazon Data Firehose and its resource as a Firehose stream, though older material and many search results still use the Kinesis Data Firehose name. A Firehose stream can also read directly from an existing Kinesis data stream as its source.

What is Amazon Data Firehose?

What drives Amazon Data Firehose pricing?

You pay for the volume ingested, with three optional add-ons: format conversion to Parquet or ORC, delivery into a VPC billed per GB processed and per Availability Zone hour, and dynamic partitioning billed on data processed, objects delivered, and JQ processing. Transfer and destination costs are separate.

Official Amazon Data Firehose pricing

How should I set buffer size and interval?

Firehose buffers incoming data until a size or time threshold is reached, so the setting trades delivery latency against object size and downstream request count. Small buffers create many small objects that raise storage request and query cost; use dynamic partitioning when the destination needs a prefix layout.

Partition streaming data in Amazon Data Firehose

Related AWS services for this plan

  • Amazon Kinesis — Use Kinesis Data Streams when consumers need replay and retention rather than one buffered delivery path.
  • Amazon Simple Storage Service — S3 is the common destination, where buffer settings decide object size and later query cost.
  • Amazon OpenSearch Service — Deliver to OpenSearch Service for search and observability, and size that domain as its own cost.
  • AWS Lambda — Attach a Lambda transform to reshape or filter records before delivery, billed as Lambda usage.

Amazon Data Firehose pricing and cost programs

Pricing model: Streaming delivery usage

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

Billing dimensions: Data ingested · Format conversion · VPC delivery · Dynamic partitioning

Programs and modes: Direct PUT · MSK sources · Format conversion · Dynamic partitioning

Ingestion source and optional transformation or delivery features add 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 Data Firehose?

  • Eliminates custom delivery workers for buffering, batching, retrying, and loading streaming data into supported analytics destinations.
  • Accepts direct writes or supported Kinesis and MSK sources and can invoke Lambda transformations, decompress records, convert formats, dynamically partition, and back up records to S3.
  • Manages delivery scaling and retries with configurable buffering, encryption, CloudWatch metrics and logs, and destination-specific error handling.

How to implement Amazon Data Firehose

  1. Choose the source, destination, latency objective, record format, compression, partitioning, transformation, and replay requirements, then verify all Region, role, endpoint, and destination-capacity prerequisites.
  2. Create the Firehose stream with least-privilege source and destination roles, KMS settings, buffering hints, Lambda transform and Glue schema where required, S3 error prefix, and source or failed-record backup.
  3. Send representative records, validate destination schema and counts, replay failed output, and alarm on data freshness, delivery failures, throttling, transformation errors, KMS errors, backup growth, and destination latency.

Amazon Data Firehose best practices

  • Configure S3 backup for source or failed records according to recovery needs, use distinct error prefixes, retain enough context to replay safely, and regularly test the replay procedure.
  • Tune buffering size and interval against latency, file size, request cost, and destination capacity; use efficient compression and columnar conversion for analytics and avoid generating large numbers of tiny files.
  • Make transformations deterministic and bounded, grant each role only required resources, encrypt interim and destination data, monitor freshness rather than ingestion alone, and scale or throttle slow destinations before backlog exceeds retention.

Amazon Data Firehose use cases and server impact

  • Streaming logs and events into S3
  • Managed delivery to Redshift and OpenSearch
  • Transforming and delivering records to Iceberg or HTTP destinations

Replaces custom streaming batcher and destination-loader services, while schemas, transformation correctness, failed-record recovery, buffering latency, and destination capacity remain customer concerns.

Official implementation resources

Commonly paired AWS services