Google today announced the beta release of Cloud Bigtable, a new managed NoSQL database on the Google Cloud Platform.
At the core of the new service is Google’s Bigtable database, which Google detailed in an academic paper in 2006. (Bigtable still plays a part in Google consumer-facing services like Gmail and Google search.) And it can be accessed through the application programming interface for HBase, an open-source implementation of Bigtable that stores and serves up data in the Hadoop open-source file system.
But Google has fine-tuned Cloud Bigtable for performance.
“The write throughput per dollar on this product is three times what the standard HBase implementation would be,” said Tom Kershaw, head of product management for storage, networking, and big data at the Google Cloud Platform.
Google created Cloud Bigtable with big companies (think petabyte-scale data sets) in mind, Kershaw told VentureBeat in an interview. Google for the past couple of years has offered the Cloud Datastore, which depends on Bigtable, but Kershaw described Cloud Datastore as “our getting-started NoSQL database.”
It’s not surprising to see Google productize its Bigtable as a new cloud service, and this time with Bigtable in the name — why not fully capitalize on technology that garnered so much fascination years ago?
The move is Google’s latest step toward the leadership position of the growing public cloud market, where Amazon Web Services currently reigns supreme and Microsoft Azure is rapidly introducing new services.
Amazon has its DynamoDB managed NoSQL database, which built on the concepts laid out in Amazon’s Dynamo paper, and now Google has Cloud Bigtable, based on Bigtable. Microsoft, for its part, made its own managed NoSQL database, Azure DocumentDB, earlier this year. Meanwhile IBM has cloud database technology Cloudant.
Google’s new offering sounded compelling to Nick Heudecker, a research director at tech analyst firm Gartner. For one thing, companies won’t need to deal with the complexity of setting up and operating HBase, Heudecker told VentureBeat in an interview. Heudecker just isn’t sure that how popular it will become among enterprises.
“It’s not clear that Google has a robust enterprise marketing and sales organization in place to truly go sell this into the enterprise,” Heudecker said.
But it’s clear Google hasn’t completely overlooked the business part of a rollout like this.
CCRi, Pythian, and Telit Wireless Solutions have already integrated their technologies with Cloud Bigtable, Google product manager Cory O’Connor wrote in a blog post on the news today. And there’s a customer to tout, too: Google has gotten digital marketing startup Qubit to move from HBase to Cloud Bigtable, according to the blog post.
And enterprises have factored in to the pricing model for the new service, which is based on chunks of throughput that customers request.
“Each node will deliver up to 10,000 queries per second and 10MB per second of throughput,” Kershaw told VentureBeat. “Enterprises can run whatever jobs they want through throughput chunks and have complete predictability in their costs.”
Time will tell how things will play out, for this service and for the Google public cloud in general.
“I’m cautiously optimistic, like I am with so many other technology rollouts,” said Heudecker, the Gartner analyst.