Google Date Centres: Inside the campus network room, routers and switches allow Google’s data centres to talk to each other. The fibre optic networks connecting Google’s sites can run at speeds that are more than 200,000 times faster than a typical home internet connection. The fibre cables run along the yellow cable trays near the ceiling. Image: Google
TO THE average eye, it would seem fairly flawless — you type in a request, Google spits out an answer — but the reality, as the tech company shared this week, is far more complex.
Secrets that have never been shared outside of Google were revealed this week at an engineering conference in Silicon Valley, detailing the “insane” approach behind how its computer software answers your questions in Google Search, directs you on Google Maps, sends your emails and allows you to watch videos on YouTube, for example.
“Behind your simple page of results is a complex system, carefully crafted and tested, to support more than one-hundred billion searches each month,” Google writes in a search explainer.
It’s all thanks to one custom-built, “giant, single shared codebase” at Google, that runs through 10 different Google data centres, Engineering Manager Rachel Potzin revealed.
They call it a “single, monolithic repository model” and unlike most software companies, this one network juggles all of Google’s software, including Google DOCS,Google+ and Gmail, across its vast network. And it’s only available to a select number of “coders” within its organisation.
All the colours of the rainbow: A Google data centre in Douglas County, Georgia. Picture: Connie ZhouSource:AP
Potzin estimated the software that keeps the service intact spans a whopping 2 billion lines of code. Wired compared it to Microsoft’s Windows Operating system, dubbed “one of the most complex software tools ever built for a single computer”, and predicted it ran along some 50 million lines. Google is the equivalent of 40 times that of Microsoft.
To keep up with the rapid evolution of the internet, its engineers modify and update around 15 million codes each week, helped by the use of bots to maintain code health, and keep the search engine running smoothly.
Google Data Centre, South Carolina: these ethernet switches connect Google’s facilities network. Thanks to them, Google is able to communicate with and monitor the main controls for the cooling system in their data centre. Image: GoogleSource:Supplied
“It’s frankly enormous and without being able to prove it, I’d guess this is probably the largest single repository in use anywhere in the world. I’d be very surprised if a larger more heavily modified single repostiry exists anywhere else,” Potzin said.
“In almost eight years our repository has grown by orders of magnitude on almost every dimension.
“There were times in Google’s history where we weren’t sure if we were going to be able to sustain this level of growth.”
In this day and age it seems like a mammoth task to handle such a gigantic bulk of information, but fast-growing, global companies like Facebook are joining the bandwagon.
It’s complex, but it’s an intriguing insight into how companies of today are bracing for the internet of tomorrow, and paving the way for how we, as humans, will interact with the online of the future.