Function as a Service (FaaS)

Author

Govardhana M K
March 17, 2025

In partnership with

TechOps Examples

Hey — It's Govardhana MK 👋

Along with a use case deep dive, we identify the remote job opportunities, top news, tools, and articles in the TechOps industry.

👋 Before we begin... a big thank you to today's sponsor SUPERHUMAN AI

Find out why 1M+ professionals read Superhuman AI daily.

In 2 years you will be working for AI

Or an AI will be working for you

Here's how you can future-proof yourself:

  1. Join the Superhuman AI newsletter – read by 1M+ people at top companies

  2. Master AI tools, tutorials, and news in just 3 minutes a day

  3. Become 10X more productive using AI

Join 1,000,000+ pros at companies like Google, Meta, and Amazon that are using AI to get ahead.

IN TODAY'S EDITION

🧠 Use Case

  • Function as a Service (FaaS)

🚀 Top News

👀 Remote Jobs

📚️ Resources

📢 Reddit Threads

🛠️ TOOL OF THE DAY

xlskubectl - A spreadsheet to control your Kubernetes cluster.

🧠 USE CASE

Function as a Service (FaaS)

Many cloud engineers think Function as a Service (FaaS) just means running code without worrying about servers. That’s not wrong, but it’s not the whole picture.

FaaS changes how we build apps, making them more event driven, scalable, and cost efficient. But it also comes with trade offs.

Before we get there, we first need to understand how Serverless operates.

Think of traditional architecture as a well structured, all in one setup. Your application, security, backend logic, and database all live together in a controlled environment. It’s predictable but requires ongoing maintenance and scaling efforts.

Now, compare that to serverless architecture (which FaaS is a big part of). Instead of managing everything in a monolithic setup, you break down workloads into small, event driven functions that scale automatically.

This shift isn’t just about backend logic, it extends to API gateways, security layers, and even database triggers.

What is FaaS, Really?

FaaS is a cloud service that lets you run code in response to events without managing servers. Your function only runs when triggered, and you only pay for the execution time.

Key traits of FaaS:

  • Event driven – Functions run when something happens (HTTP requests, database updates, queue messages, etc.).

  • Stateless – Each function runs in isolation, so you need external storage for state.

  • Auto scaling – Grows from zero to thousands of executions in seconds.

  • Short execution time – Functions usually have limits (eg: AWS Lambda allows up to 15 minutes).

FaaS shines in specific situations. Here’s when it makes the most sense:

API Backends – Deploy REST or GraphQL APIs quickly.

Event Driven Workloads – Process log changes, queue messages, or IoT events.

Scheduled Jobs – Automate tasks like backups, reports, or notifications.

Glue Code & Automation – Connect different cloud services efficiently.

Parallel Processing – Run multiple independent tasks without heavy orchestration.

Security & Compliance Hooks – Implement security policies, enforce access control, and monitor events in real time.

Practical Technical Considerations:

Cold Starts & Latency – Idle functions take 100ms to several seconds to start. To mitigate:

  • Use provisioned concurrency (where available) for critical functions.

  • Optimize startup times by reducing dependencies and using lightweight runtimes.

Networking Constraints – Functions execute in ephemeral containers with no persistent connections:

  • Use database proxies (like RDS Proxy) to pool connections efficiently.

  • Prefer event driven data pipelines over direct DB queries.

Observability & Debugging – Traditional logging methods don’t work well, instead:

  • Use structured logs in external log aggregators (e.g., Loki, Datadog, CloudWatch Logs).

  • Enable distributed tracing (OpenTelemetry, AWS X-Ray, GCP Cloud Trace).

State Management – Since functions are stateless, consider:

  • Storing transient data in object stores (S3, Cloud Storage).

  • Using managed caches (Redis) for session persistence.

Execution Time & Resource Limits – Functions are not meant for long running tasks:

  • If processing exceeds time limits, use an orchestrator (Step Functions, Durable Functions).

  • For high memory workloads, leverage tiered memory settings to optimize execution.

Cost Considerations – Billing is based on execution time, memory allocation, and request count:

  • Reduce cost by right sizing memory configurations.

  • Minimize execution time with optimized runtime environments (e.g., Golang over Python for faster cold starts).

❝

If you need long running processes, persistent connections, or deep runtime control, you might want a mix of PaaS or containerized solutions.

FaaS is a powerful but requires architectural changes, monitoring strategies, and workload specific optimizations.

You may even like:

Looking to promote your company, product, service, or event to 40,000+ Cloud Native Professionals? Let's work together.