When should I use chunked transfer encoding instead of sending the full response?

Use chunked encoding when: (1) You don't know the response size upfront — such as streaming database query results or generating dynamic content. (2) You want to reduce time-to-first-byte — start sending data before the entire response is generated. (3) The response is very large and buffering it would consume too much server memory. (4) You're implementing Server-Sent Events or real-time streaming. Don't use it for small, static responses where Content-Length is known — regular responses with Content-Length are more efficient because proxies and clients can optimize caching and connection reuse.

What is the difference between chunked transfer encoding and HTTP Range requests?

Chunked transfer encoding (Transfer-Encoding: chunked) is a mechanism for the server to stream a response in pieces when the total size is unknown. The server decides how to chunk the data. Range requests (Range header + 206 Partial Content) allow the client to request specific byte ranges of a known resource. The client controls which bytes to receive. Use chunked encoding for streaming dynamic data; use Range requests for resumable downloads of static files. They solve different problems — chunked encoding is about server-side streaming, Range requests are about client-controlled partial retrieval.

How do I implement resumable file downloads with Range requests?

On the server: (1) Include Accept-Ranges: bytes in responses for downloadable files. (2) When a Range header is present, parse the byte range, create a file stream for that range, and respond with 206 Partial Content including Content-Range: bytes start-end/total headers. (3) Return 416 Range Not Satisfiable if the range is invalid. On the client: check if the partial file exists locally, get its size, and send Range: bytes=existingSize- to request the remaining bytes. Append the response to the existing file instead of overwriting it.

How do I consume a chunked/streaming response in JavaScript?

Use the Fetch API's ReadableStream: const response = await fetch(url); const reader = response.body.getReader(); const decoder = new TextDecoder(); while (true) { const { done, value } = await reader.read(); if (done) break; console.log(decoder.decode(value, { stream: true })); }. For NDJSON (newline-delimited JSON), accumulate text in a buffer and split by newlines, parsing each complete line as JSON. Remember to pass { stream: true } to TextDecoder.decode() to handle multi-byte characters that might be split across chunks.

Does chunked transfer encoding work with HTTP/2?

No — HTTP/2 does not use Transfer-Encoding: chunked. HTTP/2 has its own binary framing layer that handles streaming natively through DATA frames. Each DATA frame carries a portion of the response body, and the END_STREAM flag signals the end. If your server sends chunked encoding, HTTP/2 implementations automatically strip the chunked framing. From an application perspective, the streaming behavior is the same (you still call res.write() in Express), but the wire protocol is different. This is handled transparently by the HTTP/2 implementation.

Chunked Responses and Partial Content in APIs