Google decided to share information about the degree of expansion of its code. By the way, under the word “code” we mean complete source code for all software, which is responsible for the functioning of the force, known as Google. If you combine source files for search engines, all services on-line, business, multimedia and hardware solutions in one file, then the collection will be 2 billion lines of code.
It's a lot, or slightly? Well, fictionalized for the first generation Jurassic Park movie with 1993 years managed software in 2 million lines of code. That world is, but, fantasy, and this is reality. Even the Linux kernel code is more complex, than Jurassic Park software – about 15 million rows.
If number 2 billion is not generalizable in your imagination, Imagine, that all files, who store this code occupy 86 terabytes. Access to them every day 95% Google Teams, which develops software. it 25 thousand people around the world, who turn to him every day, average 15 thousands of code changes 1 day. To this is added 30 thousands of changes implemented by automation systems. Throughout history the code has been edited 35 million times.
Before the amendment takes effect on a global scale, it must go through the process of adoption
Source, of course, divided into files, and it is up to 9 millions. Check out, how many files are on your computer. You will immediately receive an answer to this, what a complex structure.
Ensuring the integrity of such complex code is not easy. Each, who tried to program, He knows, what are we talking about. Google developers every week improve about 15 million lines of code, which are scattered across 250 thousands of files.
For comparison, the Windows XP code was this 45 million lines of code (more than 40 times less than), and as for its optimization, there were always many doubts. В Google, to prevent confusion, bot system created. They help clean up and optimize the code., but still an indispensable factor in this process is a person.
The rate at which Google code changes, it's exponential growth, although in recent years it has succumbed to stabilization. The real explosion of code improvement came about after 2008 of the year. See also, that the number of changes, introduced during automated systems is growing much faster, than the number of changes, introduced by developers and correlates with the number of human fixes.
To effectively manage and share individual code snippets, Google has developed its own version control system, who named Piper. Part of the code, which is responsible for the key technologies of Google, and which is in no way accessible from the outside, available only to the chosen ones. The code is available to all employees, is in copies in 10 Google data centers around the world.
It's hard to imagine, what would happen, if someone made such a joke and blocked all of google. Google is more than just a search engine, today it is an important element of our life, even if you don't feel it.
Using code snippets, Google developers can create new solutions, that integrate with Google infrastructure. Seem to be, that Google is handling the code quite effectively so far. Below is the entire presentation of Rachel Potvin, control engineer, who talked about the benefits of using a monolithic data warehouse and how to make changes.
In the future, Google wants to share its experience with other players, because, eventually, and other institutions will face the challenge of managing very complex code.
