Serverless Architectures with Spring Cloud Functions and Knative

Can a native image cut your serverless bill in half? We load-tested GraalVM vs. JVM to find out the surprising performance trade-offs.

#1about 4 minutes

Understanding serverless and function-as-a-service architectures

Serverless computing is a cloud-native model defined by scaling to zero, consumption-based pricing, and simplified deployments, with FaaS being a key implementation pattern.

#2about 2 minutes

Introducing the Mandelbrot set as a computational use case

The Mandelbrot set visualization serves as a practical example for exploring function-based architectures, broken down into mathematical and image-rendering domains.

#3about 5 minutes

Choosing a technology stack for serverless Java applications

A vendor-independent serverless stack can be built using Knative for Kubernetes-based serverless capabilities, Spring Cloud Functions for the programming model, and Google Cloud Run as the managed platform.

#4about 2 minutes

Deciding between distributed and monolithic function deployments

When designing function boundaries, consider data locality and communication overhead, as in-process calls within a single service can be far more efficient than REST calls between distributed functions.

#5about 7 minutes

Load testing and comparing JVM vs GraalVM native images

Load testing reveals that GraalVM native images offer significantly faster startup times and lower resource consumption compared to the standard JVM, though with a potential trade-off in raw throughput.

#6about 5 minutes

Exploring alternatives and ideal use cases for FaaS

While FaaS excels at elastic and event-driven workloads, alternatives like traditional microservices, other languages (Go, Rust), or messaging systems may be better suited depending on the specific requirements.

#7about 2 minutes

Q&A: Using FaaS for continuous and stable traffic

FaaS can be cost-effective even for stable traffic compared to over-provisioned dedicated instances, and its functional programming model offers benefits beyond just scaling capabilities.