Brotli

Brotli
Name	Brotli
Developer	Google
Released	2015
Programming language	C, C++
Operating system	Cross-platform
Genre	Data compression
License	MIT

Brotli

Brotli is a lossless data compression algorithm and format developed to improve compression for web and archive content. It was published by Google and designed to replace or complement existing algorithms used by web servers and clients, focusing on HTTP content-coding efficiency for text and binary assets. Brotli combines a modern dictionary, context modeling, and entropy coding to achieve high compression ratios while offering tunable trade-offs among compression speed, decompression speed, and output size.

History

Brotli's development was driven by engineers and researchers at Google with roots in work on web performance and HTTP optimization alongside standards groups such as the Internet Engineering Task Force and initiatives like HTTP/2 and QUIC. Initial public releases appeared in 2015, followed by standardization of the file format and content-coding identifier through collaborative efforts involving the IETF and implementers from projects like Chromium (web browser), Mozilla Firefox, and Apache HTTP Server. Adoption accelerated when major platform vendors such as Cloudflare, Akamai Technologies, and Amazon Web Services added support, while influential applications including Google Chrome, Mozilla Firefox, Microsoft Edge, and Safari (web browser) integrated decoders. The ecosystem growth paralleled related compression advances such as Zstandard and historical algorithms like DEFLATE and LZ77 derivatives.

Design and algorithm

Brotli's format blends a static or dynamic dictionary with a context-based modeling stage and an entropy coder inspired by Huffman coding and range coding techniques. The algorithm uses backward references reminiscent of LZ77-style sliding-window schemes plus prefix codes to represent literals, lengths, and distances. A pre-shared static dictionary contains common substrings derived from corpora of web resources, similar in purpose to phrase tables used in Lempel–Ziv family experiments, enabling efficient representation of small textual tokens encountered in HTML, CSS, and JavaScript assets. Compression levels control heuristics for block splitting, context modeling depth, and search effort; higher levels employ exhaustive parsing and joint optimization strategies akin to those used in data compression research at institutions like Bell Labs and university labs. The binary file layout includes headers with metadata, symbol tables, and compressed meta-blocks organized to allow streaming decompression for protocols such as HTTP/2 and QUIC.

Implementations and libraries

Multiple open-source and proprietary implementations exist. The reference encoder/decoder was published by Google in C and C++; third-party ports and integrations appear in ecosystems including libbrotli wrappers for Node.js, bindings for Python (programming language), modules for Ruby (programming language), and plugins for Nginx, Apache HTTP Server, and HAProxy. Language-specific implementations have been contributed to runtimes such as Go (programming language), Rust (programming language), and .NET Framework/.NET Core. Cloud and CDN vendors provide server-side modules and SDKs for popular platforms like Amazon Web Services, Google Cloud Platform, and Microsoft Azure. Tooling for archive utilities and package managers integrated Brotli into systems like Debian, Fedora, and Alpine Linux. Hardware-accelerated or vectorized decoder variants have been developed in research collaborations with institutions such as ARM Holdings and academic groups.

Performance and compression characteristics

Brotli generally yields smaller output than Gzip/DEFLATE for web text resources while offering decompression speeds competitive with other modern codecs such as Zstandard and faster than higher-compression-mode bzip2. Compression ratios vary with input type: for minified JavaScript and HTML it often achieves substantial percent-size reductions relative to Gzip at comparable or greater CPU cost on encode. Decompression is optimized for client-side use, emphasizing low memory and CPU overhead to suit browsers like Google Chrome and Mozilla Firefox and mobile environments such as Android (operating system) and iOS. Tunable parameters let implementers select operational points reducing encoding latency for on-the-fly compression in CDNs like Cloudflare or maximizing compression for static asset pipelines used by platforms like GitHub and WordPress.

Use cases and adoption

Primary adoption is in HTTP content-coding for serving web assets: responses labeled with the "br" content-encoding are supported by mainstream browsers and major CDNs. Static asset pipelines in build systems used by projects hosted on GitHub and GitLab adopt Brotli for distribution of archives and packages; package managers and container registries maintained by Docker, Inc. and distributions such as Ubuntu utilize Brotli-compressed resources where bandwidth savings matter. Search and publishing platforms, content delivery networks like Akamai Technologies and Fastly, and cloud providers leverage Brotli to reduce egress costs and improve page load metrics monitored with tools from Google PageSpeed and Lighthouse. Offline uses include compression for web archives, application bundles, and data interchange formats in projects by organizations such as Mozilla Foundation and Wikimedia Foundation.

Security and limitations

Brotli, like other decompression algorithms, has a surface for implementation bugs that can lead to denial-of-service or memory-safety vulnerabilities; deployments favor hardened, up-to-date libraries maintained by projects like OpenSSL-adjacent teams and vendors. The static dictionary approach can create fingerprinting vectors when small responses compress to predictable outputs, raising privacy considerations discussed in contexts involving TLS telemetry and HTTP/2 multiplexing. Resource consumption during high-compression-level encoding can be significant for server-side CPU and memory, prompting conservative defaults in CDNs and web servers. Interoperability limitations arise with legacy clients that lack "br" support; therefore many servers negotiate fallback encodings such as gzip to maintain compatibility with older user agents like legacy builds of Internet Explorer and early Safari (web browser) versions.

Category:Data compression