Vercel has unveiled a significant evolution in serverless function computing with its new Fluid compute model—a transformative step that redefines how web applications scale and perform. Fluid compute seamlessly combines the efficiency of serverless architecture with the power and flexibility of dedicated servers, enabling your applications to scale dynamically and execute multiple requests concurrently with minimal latency.
Fluid compute optimizes resource utilization by triggering compute only when needed and reusing existing resources before spinning up new ones. This means your functions can now handle a higher number of concurrent requests without the traditional pitfalls of cold starts. Enhanced with a Rust-based runtime and bytecode caching, Fluid compute dramatically accelerates function initialization and response times, ensuring that even when a cold start occurs, its still is fast.
In addition to faster processing, it introduces smarter scaling. By leveraging warm compute and a many-to-one execution model, it not only supports many of concurrent invocations but also reduces wasted compute cycles. This intelligent scaling translates into cost savings of up to 85%*.
* I personally think 25% to 50% is a more realistic number. But it always depends on your project(s) of course.
Traditional serverless
Fluid computing
Furthermore, it extends the lifecycle of a function’s invocation. With features like waitUntil, background tasks such as logging, analytics, and database updates can continue processing after the client receives its response. This innovative approach is particularly advantageous for complex applications, including those in e-commerce and AI, where post-response processing is critical for optimal performance.
Watch the announcement video below to learn more about how Fluid compute can transform your web application infrastructure.
So you're as excited as I am, be sure to enable it on your project(s). Go to Vercel.com and log in. Select the project you want to enable it for and go to Settings. Scroll down to Functions. Just flip the switch to enable it. Only thing left is to trigger a new deployment to enable the settings.
Happy codin'!
Source: