Brotli - Google’s new web data compression algorithm

As websites and online services become harder every year, the need for data compression on the web is growing. For this reason, Google has released a new data compression algorithm for websites - Brotli, which means “small loaf of bread” in Swiss German. The algorithm is already available to a wide audience on GitHub .

Brotli is open source and allows you to compress data 20-26% more efficiently than its predecessor from Google, the Zopfli algorithm (also a bakery product from Switzerland, most similar in appearance to our braid bun). Both algorithms have a simple and simple goal - to help load web pages faster.

Google’s development allows lossless data compression using a combination of LZ77 algorithms and Huffman coding, which puts Brotli on a par with the best general data compression methods currently available. At the same time, Brotli works better than LZMA and bzip2, and according to Google, the new algorithm can be compared with Deflate ZLIB in terms of speed.

The issue of data compression is especially acute for mobile users and Google hopes that the technology developed there will be universally integrated into web browsers in the future, which will allow loading pages faster. This in turn will save battery power and reduce web traffic.

The question arises as to how exactly Google achieved such data compression, to which they give the following answer (in the original, in order to avoid divergence in translation):

The higher data density is achieved by a 2nd order context modeling, re-use of entropy codes, larger memory window of past data and joint distribution codes.

Here you can find a comparison of the effectiveness of Brotli and other algorithms, which was prepared by Google (the link was provided by the user evnuh ).