Supersize it: how Motorola transformed its data warehousing and analytics with Google Cloud Platform

It’s no secret that we use big data and analytics extensively within Google, but extensive and sophisticated use of big data and analytics by enterprise organizations remains somewhat of a rarity.

One exception is Motorola Mobility, which Google owned from 2012 to 2014, and is now a Lenovo company. Motorola had always collected and analyzed a lot of data about its devices, but during its short stint as a Google company, began using analytics tools that gave it new levels of insight. Today, over 40% of Motorola employees perform north of 90,000 queries per day against over four petabytes of data stored in Google Cloud Storage and accessed with Google BigQuery.

Back in the day, Motorola had been doing what a lot of organizations are still doing: running in-house Hadoop and MapReduce clusters. By moving various analytics applications to Google App Engine and BigQuery, Motorola experienced immediate operational benefits. Before, Hadoop and MapReduce jobs frequently stalled because its in-house data centers didn't have sufficient elasticity to meet burst demand. In contrast, BigQuery jobs enjoy the virtually infinite scalability of Google Cloud Platform, and don’t require operators to intervene and rescale the cluster to accommodate spikes in demand.

Less operational overhead enabled Motorola to collect more data, and to think about the data it was gathering in new and creative ways. Engineers had always gathered data related to device stability and battery life, but began adding to its data pipelines, and granting access to it to other stakeholders within the organization. For example, support engineers can quickly join BigQuery tables about device stability and other device check-in data to help customers troubleshoot issues.

But expanding the amount of data it gathers and the number of people that have access to it raises questions of its own. To help more users directly use the data, Motorola has had to step up internal training and tools to make analysis easier. And the ease with which it’s possible to gain access — and share and transform — data with BigQuery, can lead to data quality problems. While Google Cloud Platform has democratized access to the data, it’s also highlighted the need for a well trained and conscientious workforce that uses its resources wisely and judiciously.

Going forward, that could mean exploring new data analytics tools. For example, instead of performing simple table joins with BigQuery, is there a case to be made for using Cloud Dataflow, Google’s managed batch and streaming data service? Could Google Cloud Pub/Sub emerge as a way to impose discipline on the sheer volume of tables and schemas that Motorola users create? We don’t know the answer, but look forward to watching Motorola as it takes its analytics practice to the next level.