Amazon SQS

Amazon SQS is a managed cloud message queuing service for microservices, distributed systems, and serverless applications. It lets software components send, store, and receive messages so they can operate independently and respond to events. Standard queues use at-least-once delivery and best-effort ordering; FIFO queues preserve message order without introducing duplicates. Standard queues support nearly unlimited transactions per second per API action, while FIFO high-throughput mode supports up to 70,000 messages per second without batching.

Messages can contain up to 256 KB of payload, and larger messages can use the Java Extended Client Library with Amazon S3. SQS supports batches of up to 10 messages and retention of up to 14 days. Dead Letter Queues can receive messages that exceed a queue’s maximum receive count and must match the source queue type. AWS lists integrations with Redshift, DynamoDB, RDS, EC2, ECS, Lambda, and S3, plus fanout with SNS. Server-side encryption uses AWS KMS keys. The Free Tier includes 1 million requests per month; S3 storage, KMS calls, and some data transfers can incur separate charges.

SQS provides queue options for different delivery and ordering needs, along with integrations across AWS services. The Free Tier includes 1 million monthly requests, but some related storage, encryption, and transfer use may be charged separately.

SQS suits organizations building microservices, distributed systems, or serverless applications, including startups, enterprises, nonprofits, and governments. It is useful when application components need to exchange messages without being directly coupled.

Amazon SQS is a managed cloud queue for applications whose components need to exchange work without waiting on one another. It is best suited to AWS-centered microservice, distributed-system, and serverless teams that want throughput choices without operating queue servers. The main decision is whether your workload can accept Standard queue delivery behavior or requires FIFO ordering and exactly-once processing.

Overview

SQS sits between producers and consumers: one component sends a message, SQS stores it, and another component receives it later. That separation supports event-driven designs and lets services continue working independently when another service is busy or temporarily unavailable. AWS has offered SQS since 2006, and its cloud deployment removes the need to install and maintain a messaging server.

Standard queues provide at-least-once delivery and best-effort ordering, so they are the practical choice when volume matters more than sequence. FIFO queues preserve order and provide exactly-once processing through ordered message groups. Selecting the queue type is therefore an architecture decision, not merely a performance setting.

Key features

Throughput, batching, and retention

Standard queues support nearly unlimited transactions per second per API action, making them appropriate for workloads with highly variable or very high demand. FIFO high-throughput mode reaches up to 70,000 messages per second without batching while retaining order. Batches hold up to 10 messages, which can make high-volume producers more efficient. Messages remain available for up to 14 days, enough for delayed processing but not a substitute for long-term event storage.

Payloads and oversized messages

The normal payload limit is 256 KB. Applications with larger content can use the Java Extended Client Library with Amazon S3, but that adds S3 storage and its associated billing to the design. The published maximum message size is 1 MB, so SQS is better for commands and event metadata than for large documents or media.

Failure handling and integrations

Dead Letter Queues can receive messages after a configured maximum receive count is exceeded, giving operators a place to isolate repeatedly failing work. The dead-letter queue must use the same queue type as its source, which matters when designing Standard and FIFO workflows. AWS lists integrations with Redshift, DynamoDB, RDS, EC2, ECS, Lambda, and S3, and describes fanout patterns with SNS. Those connections make SQS particularly practical inside an existing AWS estate.

Security and protocols

Server-side encryption uses AWS KMS keys: SQS encrypts messages on receipt and decrypts them for authorized consumers. KMS calls can incur separate charges. Clients can use AWS JSON, AWS Query, HTTPS, and TLS protocols. Cross-Region or internet data transfer may also add charges, so a low queue request bill does not necessarily represent the full cost of a distributed deployment.

Pricing

SQS uses usage-based pricing rather than seats. The Amazon SQS Free Tier costs nothing and covers 1 million requests per month, calculated monthly across Regions except GovCloud; unused allowance does not roll over. This is a useful entry point for prototypes and low-volume services, but request growth, KMS activity, S3 storage, and data transfer can create charges beyond the free allowance.

Standard queues have custom pricing with usage-based billing, at-least-once delivery, and best-effort ordering. They fit throughput-first systems that do not need strict sequencing. FIFO queues also have custom pricing with usage-based billing, exactly-once processing, and ordered message groups. Choose FIFO when duplicate prevention and order justify its specialized behavior; otherwise Standard is the simpler throughput option. There is no separate seat limit or fixed trial term stated for these plans.

Platforms

SQS is delivered as a cloud service through APIs and the web. It supports AWS JSON, AWS Query, HTTPS, and TLS, making it suitable for services and applications that can make AWS API calls rather than requiring a self-hosted broker.

Who it's for

SQS is a strong fit for startups, enterprises, nonprofits, and governments building microservices, distributed systems, or serverless applications on AWS. It is especially compelling when Lambda, ECS, EC2, S3, or other listed AWS services already form the application stack. Teams needing self-hosting, broker-style administration, or payloads larger than the queue limits should look elsewhere.

Pros and cons

Pros

• Two meaningful delivery models: Standard maximizes throughput, while FIFO addresses ordered, exactly-once workflows.
• High scale: Standard offers nearly unlimited transactions per second per API action, and FIFO high-throughput mode reaches 70,000 messages per second without batching.
• Useful resilience tools: 14-day retention, batches of 10, and dead-letter queues support delayed work and failure isolation.
• Broad AWS fit: Integrations across compute, database, analytics, storage, and SNS fanout reduce plumbing in AWS architectures.
• Accessible starting cost: The 1-million-request monthly Free Tier supports experimentation and small services.

Cons

• Standard ordering is not strict: Best-effort ordering and at-least-once delivery require consumers to tolerate reordering or duplicates.
• Small native payloads: The 256 KB limit pushes larger messages toward S3 and the Java Extended Client Library.
• Cost can extend beyond requests: KMS calls, S3 storage, and cross-Region or internet transfer may be billed separately.
• Cloud-only operation: SQS does not suit organizations that require a self-hosted deployment.

Alternatives

Browse Message Queue (MQ) Software and Message Broker Software for broader comparisons.

• EMQX is a freemium, self-hosted-capable option with a Serverless plan at no cost and up to 1,000 connections, making it worth considering for MQTT-oriented deployments.
• HiveMQ MQTT Broker offers MQTT-focused plans, including a Lab tier capped at 100 connections and 100 messages per second.
• Aiven for Apache Kafka suits Kafka workloads; its Free plan provides 250 KiB/s throughput, 3 days of retention, 5 topics, and 2 partitions per topic.
• Confluent Cloud is an alternative for managed Kafka with a free Basic plan and autoscaling resources.
• Google Cloud Pub/Sub is the natural alternative for Google Cloud teams, with 10 GiB of free throughput before its $40 monthly pay-as-you-go tier.
• Amazon MQ is a paid managed messaging service with a six-month AWS Free Tier for eligible new customers.
• Apache ActiveMQ is free, open-source, and self-hosted under the Apache License 2.0.
• Apache Kafka is free to download and self-host, without formal commercial support from the Apache Software Foundation.

Verdict

Choose Amazon SQS when you need managed queues inside an AWS application and want a clear trade-off between Standard throughput and FIFO ordering. Its integrations, dead-letter handling, and generous request-based Free Tier make it a dependable foundation for decoupled services. Look elsewhere if self-hosting, strict ordering on every queue, or large native message payloads are non-negotiable.