Back in 2012, Google unveiled Spanner — a globe-spanning database technology, the first of its kind, designed to keep the search advertising giant’s massive online operations in sync all over the world.
Nowadays, Spanner is the database behind Google services like Gmail, Google Photos, and AdWords.
But getting Spanner off the ground over the course of five years required Google to rethink how it built systems.
The most dramatic example is “TrueTime,” a system for making sure that every single one of its massive data centres had the exact same time on its clocks, down to the picosecond, so that data flows in and out of Spanner in the right order. To get TrueTime running, Google had to build antennae on top of every one of its 15 data centres, so multiple GPS satellites could cross-check the time with each other.
Now, for the first time ever, Google is opening Spanner up to the outside world with Cloud Spanner, the newest arrow in Google Cloud’s quiver as it takes aim at the $US10 billion Amazon Web Services juggernaut, and the fast-growing Microsoft Azure cloud.
Google Cloud customers can now basically rent out Spanner for their own apps and services, using the same technology that Google uses in-house.
“CloudSpanner is literally a unique service,” says Google Cloud Lead Product Manager for Databases Dominic Preuss. “There is literally no other way to do this.”
Like Oracle, MySQL, IBM DB2, or any number of other database software products, Cloud Spanner uses the SQL language for “querying,” or getting information from, the system. That means it’s a familiar experience for anybody coming from those kinds of databases, which would be most customers.
But like newer kinds of databases, including trendy systems like the Facebook-created Cassandra, it’s designed to handle massive amounts of information.
And, even for those customers that don’t necessarily need their databases to hold that much data, Cloud Spanner is still hosted from those very same Google Cloud data centres, hardened against hardware failure, cyberattack, and everything in between.
From Preuss’ perspective, that alone is a huge selling point: Google handles the hard part, including operating the servers; you just reap the benefits of their tech.
The competitive edge
Google really believes that Cloud Spanner is a better way of building databases, as proven by its own usage — “just because we’re different doesn’t make us a [fragile] snowflake,” jokes Cloud Spanner Product Manager Deepti Srivastava.
Still, it requires some retooling and some effort to make Cloud Spanner work with existing apps, and Srivastava acknowledges that not every customer will want to (or be able to) put in the effort to move their stuff over. For those customers, Google offers Cloud SQL, a product that’s much more like traditional database software.
In the bigger picture, Srivastava says that with today’s news, Google Cloud is a “one-stop” data shop: Cloud SQL for traditional customers; Cloud Spanner for the adventurous and those building new apps; and the BigQuery data analysis engine for actually working with the data stored therein.
In a competitive sense, Oracle, Amazon, and Microsoft all specifically pitch their cloud computing platforms as the best place to run databases — with Oracle’s cloud optimised for Oracle databases and Microsoft’s optimised for its own database products.
But Preuss would point out that none of them have Google’s experience with dealing with massive amounts of data, at huge scales, and working with them in real time.
“Google at its core is a data company,” says Preuss.
Business Insider Emails & Alerts
Site highlights each day to your inbox.