Extreme optimization that makes WebStreams 10x faster

During profiling of Next.js server rendering, it was confirmed that WebStreams itself is a performance bottleneck. Even though the data is already in the buffer, each call to reader.read() triggers object allocation, a Promise chain, and a move to the microtask queue, causing significant overhead.

In fact, with 1KB chunks, the native WebStream's pipeThrough performance was only 630 MB/s, whereas Node.js's pipeline() recorded about 7,900 MB/s, showing a performance gap of more than 12x.

To solve this, the fast-webstreams library was developed, which maintains the WHATWG standard API while internally using optimized Node.js streams. The main optimization approaches are as follows:

• Pipeline optimization: When chaining pipeThrough and pipeTo, instead of executing immediately, the upstream links are recorded and then consolidated into a single pipeline() call at the end, eliminating Promise overhead. This achieved a performance of about 6,200 MB/s.
• Fast-path read: When reader.read() is called and data is available, it is processed synchronously via Node.js's read(), minimizing event loop round-trips and object allocation. This approach shows a speed of about 12,400 MB/s.

This optimization dramatically improves performance especially in environments that use specific byte stream patterns, such as React Server Components.