Google is 2 billion lines of code

The publication Wired drew attention to the speech of Google employee Rachel Potvin "The Motivation for a Monolithic Codebase", which was held at the conference in Silicon Valley. In her report, she estimated the number of lines of code that is responsible for the operation of all Google Internet services: it turned out that the number is approximately 2 billion. If you make an incorrect comparison and take into account that Windows contains about 50 million lines of code, it turns out that since 1998 Google has managed to write 40 Microsoft operating systems, which has been developed since 1985. Moreover, this entire “Google code” is in a single repository, which is used daily by 25,000 employees of the search giant.

Rachel noted that this principle of source code storage allows Google developers "... to feel unusual freedom in using and combining the code of other projects." The only existing limitation is access to the code that implements Google's PageRank ranking algorithms, which are the foundation of a business that is critical for the search giant. These files have access only to dedicated employees. In general, Google’s own VCS, called Piper, is used to manage the code on Google, which in turn relies on a serious infrastructure consisting of 10 data centers.

Curious statistics given Rachel. According to her, Piper manages 85 terabytes of Google-code data, with 25,000 developers making about 40,000 commits daily. Thus, 250,000 files and 15 million lines of code are modified every week. Compared to Linux, which totals about 40,000 files, the work of Google programmers looks fantastic. Rachel also noted that Google and Facebook are now working together on a new open source VCS that can be used on projects of any scale comparable even to Google itself. . Interestingly, the foundation of the future VCS is not fashionable among git developers, but Mercurial, which they are trying to scale to the level that both Internet giants have to deal with.