Clase grupal: Hablando desde nuestra experiencia

What has been a big challenge in working with serverless on AWS?

When we talk about serverless on AWS, it initially seems like a perfect solution because of its ability to scale. However, one of the biggest headaches is IP address management. In one particular experience, we were required to host an application within a network with a very limited range of IPs. The serverless nature of AWS Lambda, which allows for almost infinite scaling, resulted in a network collapse due to a lack of available IPs. This type of problem underscores the importance of understanding the limits of the cloud infrastructure and making sure to properly plan for scalability.

When should you not use serverless for an application?

Although serverless offers significant advantages, it is not always the best choice. In applications that require constant processing of large volumes of data, such as ETL tasks, serverless may not be efficient because of the way it is billed on a per-use basis. These processes may demand more resources if performed with Lambdas, which is not cost-effective compared to dedicated S2 instances that could execute such tasks faster. Therefore, in intensive and continuous processing applications, investing in dedicated instances could be a more economical and efficient decision.

Where to look for reliable information on serverless?

Curating a list of reliable sources to keep up to date on serverless is essential for a developer. Some recommended resources include:

• Serverless First and Serverless Land: Web sites focused on serverless technologies.
• Medium: A platform full of technology articles that often hosts current discussions and first-person experiences.
• Twitter: Follow industry experts who frequently share news and updates.
• Platzi: Courses and forums that offer up-to-date information and discussion forums on serverless best practices and developments.

These sources provide not only theories, but also practical applications and use cases that enrich technical knowledge.

How to handle a serverless application in production?

For a serverless application to work optimally in production, it is crucial to integrate multiple tools and processes:

• GitHub Actions: Allow to manage deployments to different environments (Dev, QA, UAT, Prod).
• Monitoring and Observability: Tools to analyze performance.
• CICD (Continuous Integration and Delivery): Provide automatic delivery flows and updates without visible interruptions for the user.
• Infrastructure best practices: These are key in both traditional and modern paradigms and should be customized for serverless.

In addition, it is critical to evaluate specific serverless solutions to ensure compatibility with general industry practices.

Are there other tools that complement the serverless ecosystem?

Serverless does not replace existing technologies such as Kubernetes, but can coexist with them for a robust ecosystem. For example, Kubernetes handles containers that might be ideal for microservices that require constant scale, while serverless can handle services that react to sporadic events. This coexistence between tools represents an evolutionary advance, rather than a replacement, highlighting the importance of choosing the best of both technologies depending on the particular use case.

How to compare serverless framework with Terraform or CloudFormation?

Each tool in the cloud ecosystem has its strengths and ideal applications. Serverless framework allows integrating code development with the infrastructure directly, decreasing the distance between dev and ops. However, Terraform is another solid option for provisioning and managing more traditional cloud infrastructure, although it is not ideal for managing serverless code directly. In a mixed environment, it can be useful to combine serverless framework for managing Lambda functions and Terraform for broader infrastructure architectures, such as Kubernetes clusters. This synergy allows you to leverage the strengths of each tool.

William, we invite developers to experiment with different tools without getting pigeonholed, choosing those that best suit their needs even though serverless is tempting. Explore, test and keep learning!