Skip to content

Serverless on AWS: why SMBs are adopting it

How serverless architecture on AWS helps SMBs reduce infrastructure costs and focus on their core business.

Updated on 18 November 2024

Infrastructure that adapts to your actual usage

Serverless architecture on AWS lets you run code without managing servers. You only pay for the actual execution time of your functions, down to the millisecond. For SMBs, this means zero infrastructure cost when there is no activity and automatic scaling during demand peaks. AWS Lambda, the flagship serverless service, processes billions of requests per month for businesses of all sizes.

How serverless works

In a traditional architecture, you rent an EC2 server that runs 24/7, whether you use it or not. With serverless, you deploy functions that execute only when called. AWS manages the underlying infrastructure: provisioning, scaling, patching and monitoring.

AWS Lambda is the central service of serverless architecture. You write a function in the language of your choice (Python, Node.js, Java, Go) and Lambda executes it in response to an event: HTTP request, file upload, queue message or scheduled trigger.

API Gateway exposes your Lambda functions as REST or WebSocket APIs accessible from the web. DynamoDB provides a serverless NoSQL database that automatically adapts to load. S3 stores files. These services combine to create complete applications without any server to manage.

Concrete use cases for SMBs

Form processing

A contact form on your website triggers a Lambda function that validates data, sends a confirmation email to the customer and stores the request in DynamoDB. The cost is a few cents per month for hundreds of submissions.

Image processing

When a user uploads an image to S3, a Lambda function automatically resizes it into multiple formats (thumbnail, medium, high resolution). This on-demand processing avoids maintaining a dedicated image processing server.

Scheduled tasks

Recurring tasks like report generation, data synchronization or scheduled newsletter sending execute via EventBridge triggers. You only pay for the few seconds of execution time per task.

Mobile application APIs

A serverless API with Lambda and API Gateway provides the backend for your mobile application. The infrastructure automatically adapts to the number of users, from 10 to 10,000, without manual intervention.

The cost model

AWS Lambda charges $0.20 per million requests and $0.0000166667 per GB-second of compute. In practice, an SMB processing a few thousand requests per day pays less than 1 euro per month for Lambda. The AWS free tier includes 1 million requests and 400,000 GB-seconds per month, which covers many use cases.

DynamoDB charges based on consumed capacity. On-demand mode suits unpredictable workloads. API Gateway charges $3.50 per million requests. A complete serverless architecture for an SMB typically costs between 5 and 50 euros per month.

Benefits for SMBs

Serverless eliminates infrastructure maintenance burden. No server to patch, no operating system to update, no capacity to plan. Your team focuses on developing business features rather than system administration.

Automatic scaling protects your business against unexpected traffic spikes. A marketing campaign that generates a sudden influx of visitors is absorbed without intervention. Conversely, quiet periods cost nothing.

Serverless is an approach particularly suited to SMBs that want to modernize their infrastructure without investing in server administration. LCMH designs serverless architectures on AWS for businesses in Alsace.

For more on optimizing your cloud costs, read our article on 7 AWS cost optimization levers.


Sources

  1. AWS, AWS Lambda. aws.amazon.com/lambda
  2. AWS, Serverless on AWS. aws.amazon.com/serverless
  3. AWS, AWS Lambda Pricing. aws.amazon.com/lambda/pricing

Frequently asked questions

Serverless est-il moins cher que les serveurs EC2 ?
Pour les charges de travail intermittentes ou à faible volume, le serverless est nettement moins cher car vous ne payez que le temps d'exécution réel. Pour les charges constantes et élevées, EC2 avec Savings Plans peut être plus économique. Le point de bascule se situe généralement autour de 30% d'utilisation continue.
Quelles sont les limites du serverless ?
Les principales limites sont le temps d'exécution maximum (15 minutes pour Lambda), le cold start (latence au premier appel après une période d'inactivité) et la complexité du debugging distribué. Ces limites conviennent à la majorité des cas d'usage des PME.
Faut-il un développeur pour mettre en place du serverless ?
La implementation initiale nécessite des compétences en développement et en configuration AWS. Une fois en place, les fonctions serverless ne demandent aucune maintenance d'infrastructure. Un consultant AWS certifié peut développer et déployer ces solutions turnkey.

Related Articles