February 17, 2017
Google’s Spanner, developed a decade ago, created a way to store information across millions of machines in a multitude of data centers around the world. Despite its global reach, Spanner behaves as if it operates in a single location, meaning it can reliably replicate and change data without contradicting actions taken at a different location, and retrieve copies if one of the centers goes down. Since its creation, Spanner has become the foundation for 2,000 Google services including Gmail and AdWords. Now, Google plans to unveil Spanner to everyone as a cloud computing service.
Wired opines that, “Google believes this can provide some added leverage in its battle with Microsoft and Amazon for supremacy in the increasingly important cloud computing market.” Stanford University computer science assistant professor Peter Bailis agrees, noting, “If they offer it, people will want it, and people will use it.”
But those who disagree point out that few companies have the same need as Google for millions of machines in different global locales to communicate in a way that “transactions in one part of the world lined up with transactions in another,” giving an accurate global snapshot of operations to “seamlessly replicate data cross regions or quickly retrieve replicated data when it was needed.”
Using GPS receivers and shuttling time readings to master servers in each worldwide data center, Spanner settles on a common time, coming up with a reliable time-keeping service that Google calls TrueTime.
Former Google engineer Peter Mattis, whose startup CockroachDB is trying to build an open source version of Spanner, calls it “faster-than-light coordination between two places.” But Cockroach won’t be able to replicate TrueTime, because only Google (along with a small number of other global tech companies) has such a massive global infrastructure.
By opening up Spanner to the outside world, Google hopes to “convince customers that Spanner provides an easier way of running a global business,” especially to replicate data across multiple regions to avoid the pitfalls of outages. Right now, JDA Software Group, a company that “helps businesses oversee their supply chains,” is testing Spanner, and JDA group vice president John Savari reports that, “The volume of data — and velocity with which that data is coming at us — is amplifying significantly.”
Spanner might also be a good fit for financial markets; Google says it is already in discussions with “large financial institutions,” as banks become more accepting of doing business in the cloud.