Broadening its effort to challenge Amazon Web Services, Google on Wednesday plans to introduce Google Cloud Bigtable, a managed cloud database service based on the technology Google uses for its own applications such as Google Search and Google Maps.
Google began developing Bigtable over a decade ago as a high-performance distributed data storage system. Now the company is offering its infrastructure for rent.
"As a system, it's designed not just as a data warehouse," said Miles Ward, global head of solutions for Google's cloud platform, in a phone interview. "This is a real-time, high performance, transactional database."
Bigtable "is one of the foundational technologies for all of Google," said Ward. It is a NoSQL database, which means it's suitable for real-time transactions, analytics, and reporting. Ward contends that it offers roughly twice the performance for the price of comparable systems like Amazon DynamoDB, Cassandra, or HBase running on standalone compute infrastructure.
Google Cloud Bigtable works with the open source HBase API and can be integrated with a variety of applications like Google BigQuery, Google Cloud Dataflow, Hadoop, Spark, and Storm.
"As a utility for customers," said Ward, "we're able to unlock workloads that were previously too expensive to do."
Ward said Cloud Bigtable is particularly well-suited for Internet of Things applications, where companies must deal with a large volume of sensor data rapidly. A large hotel with connected door locks, he notes, would want to provide responses to access requests in milliseconds rather than, say, a week later in a batch audit process.
Google is also working with partners to implement industry-specific Cloud Bigtable applications. Sunguard, for example, is developing a financial audit product that can handle 2.5 million trade messages per second.
Qubit, a UK-based digital marketing firm, has been testing Google Cloud Bigtable.
Qubit is "the love child of analytics and ad tech," said Emre Baran, a former Google product manager and current CTO of Qubit, in a phone interview. Qubit provides a software platform that brings analytics, business intelligence, personalization, and A/B testing together.
To help e-commerce clients understand customer behavior, Qubit's software stores the information generated through online interactions as a raw event. It then processes that data on demand to provide what amounts to just-in-time online personalization.
That requires rapid storage, data lookups, and processing. Baran said Google's decision to integrate the HBase API made switching to Cloud Bigtable as simple as changing a few lines in a configuration file. His company has been using HBase, but is replacing parts of its HBase installation because Google Cloud Bigtable is easier to maintain, and it simplifies regional provisioning.
"Google frees us from that burden because the storage is just whatever data you put in there," said Baran.
Google Cloud Bigtable is interesting because setting up and deploying database software can be a challenge, and Google is making that easier, said Nick Heudecker, research director of information management at Gartner, in a phone interview. But he stressed that it's too early to tell how the service will affect the database market because it's unclear whether Google can win over enterprise IT shops.
"The challenge Google faces is that they're kind of an unknown quantity in the enterprise," said Heudecker.