What is Serverless Computing?

Serverless computing allows developers to deploy code without the need to manage servers. For instance, using a predefined function in Python, we can upload it to the cloud, where it operates within a sandbox environment provided by the cloud service. The specifics of how the function executes, including container reuse, are determined by the service provider and can vary widely.

Although termed "serverless," this model does not eliminate servers entirely; instead, it utilizes numerous decentralized services that automatically deploy in response to specific events.

While some may draw parallels between serverless and microservices, they are distinct concepts. Microservices typically encompass broader functionalities and resources, often requiring independent databases and messaging systems. In contrast, a serverless function is a concise piece of code designed to execute a single task triggered by an event. Depending on the architecture, a microservice may correspond to one or several functions.

Instead of writing traditional frameworks like Flask or Django, developers rely on the cloud provider's runtime environment. This platform autonomously manages aspects such as environment reuse and load balancing.

With serverless architecture, expenses are incurred only during code execution, similar to a pay-as-you-go model. This means that rather than maintaining a server around the clock, developers can use lambda functions that activate only when requested. This is especially advantageous for prototyping, as services like AWS Lambda offer up to a million free requests each month.

State Management and Scalability

In serverless functions, state is not retained; global variables and persistent storage on local disks are typically not options. Instead, developers integrate external databases, caches, and storage solutions. This design promotes seamless horizontal scaling in response to fluctuating load and request volumes, allowing developers to focus on coding and business logic rather than infrastructure concerns.

Here, the code serves as a connector rather than the core of the application. Applications transform user data into actionable insights and store them accordingly. In a serverless environment, the provided infrastructure links various services, enabling developers to create functions that manage data flow between web servers and databases. The result is a network of interconnected functions, where the application is more about orchestrating these components than about the code itself.

Challenges of Serverless Computing

Despite its advantages, serverless architecture does have drawbacks. One significant concern is vendor lock-in; functions created for AWS, for example, may not easily transfer to Google Cloud due to differing service ecosystems. Although Python remains consistent across platforms, the integration with various services—databases, messaging systems, logging—can vary significantly between providers.

Additionally, relying on third-party services may lead to reduced control over the system, complicating understanding and management. This can impose limitations on application capabilities.

Another downside is the phenomenon of "cold starts." If a function has not been invoked recently, it may take several seconds to initialize when called, which could adversely affect application performance.

Applications Suited for Serverless

Serverless architecture is not universally applicable. It does not represent a technological panacea or the ultimate evolution of software design. Instead, it serves specific niches effectively. Long-running tasks are unsuitable due to vendor-imposed execution time limits (e.g., AWS allows a maximum of 15 minutes). Similarly, complex applications with extensive dependencies may face challenges since developers lack control over the underlying operating system.

AWS Lambda: A Leading FaaS Provider

AWS Lambda is a prominent player in the FaaS market, supporting numerous programming languages (including Ruby, Python, Go, NodeJS, C#, and Java) and a wide array of services to tackle complex tasks. It automatically scales based on incoming requests, ensuring that costs are incurred only during function execution. If a function is idle, no fees apply. AWS also provides a free tier, albeit with certain limitations.

However, pricing can be unpredictable, with costs arising from various internal transfers and interactions with other services. Users must remain vigilant to avoid surprises in billing.

Alternatives to AWS Lambda include Azure Functions, GCP Functions, and Yandex Functions. The simplest way to initiate a lambda function is through the AWS management console, which even includes a built-in code editor.

Deployment Process

Creating a lambda function involves defining an event that contains request data and a context that holds runtime information. AWS does not enforce the use of additional abstractions, allowing developers to work with native Python.

To connect the function to the internet, developers can use AWS Gateway to create an endpoint with minimal effort. All incoming requests to this endpoint trigger the lambda function. However, as more resources are added, the complexity of managing these connections increases.

Is it User-Friendly?

Not particularly. Even straightforward tasks can feel cumbersome, necessitating careful consideration and management.

The Need for Frameworks

To simplify deployment and management, various frameworks exist that automate processes, minimizing direct interaction with the AWS console and reducing the need to handle zip files. Examples include the Serverless Framework and Zappa.

Recommended Readings

• Learning AWS Serverless Computing
• Learning Serverless

Thank you for reading!

If you have any questions or thoughts, feel free to leave a comment below to initiate engaging discussions! Also, visit my blog or connect with me on Twitter or subscribe to my Telegram channel.

Plan your best!

The first video titled "We stopped using serverless. The results are insane." discusses the unexpected outcomes and insights gained after moving away from serverless architecture.

The second video titled "That's It, I'm Done With Serverless*" reflects on the challenges faced with serverless technologies and the decision to abandon them.