As Google aims to bolster its enterprise business, it's making improvements to BigQuery, its data analysis service, to make it more compatible with traditional big data workflows.
The cloud-based service now supports standard SQL, Google announced Thursday. In conjunction with this new dialect, Google has added features such as more advanced query planning and optimization, allowing for complex subqueries in any clause of a SQL statement. Additionally, BigQuery now supports ThetaJOIN, as well as a broader type system that includes dates, times, arrays and structs.
Google also announced that it's making Cloud IAM, its new identity and access management feature, available for BigQuery in Beta. This should fully automate managing permissions for BigQuery projects, allowing administrators to create detailed security policies.
Lastly, Google introduced time-based table partitioning for BigQuery, making it easier to manage data and write queries spanning multiple days, months or years. Previously, one would have to manually partition tables to scan specific time frames.
The new features follow other enhancements to Google's analytics offerings, such as the BigQuery integration with Google Drive, and the rollout of its free Data Studio visualization toolset. More generally, Google has been working on integrating its tools. For instance, a developer could use the newly enhanced Firebase to build, manage and monetize apps, and then could export Firebase Analytics data to BigQuery.
International entertainment company Kabam is one customer that's made use of BigQuery to gather insights about its free-to-play video games. The company first turned to Google in 2015 after it launched Marvel: Contest of Champions. The game was generating a terabyte of data a day, and Kabam's self-hosted data warehouse couldn't keep up. Now, the company can focus on analyzing the data rather than scaling the necessary technology, Kabam CTO Jeff Howell said.
"We've had queries that were taking hours before and now are taking seconds or minutes," he said.
Google, Howell said, offers the flexibility, as well as the scale and performance needed when collecting the vast troves of diverse data sets required for gaming.
"Games is a really interesting data problem because every game has very different data requirements," he said. Rather than just tracking clicks, "a game has interactions with players, analytics about fighting, the types of moves they'd use, the characters and what level the're at -- it's a rich data set that is really cool but comes with a lot of technical challenges."