Google has introduced a service for storing large amounts of data online, potentially enabling organizations to execute big data analysis as a cloud service.
The offering, called Google Cloud Bigtable, "is based on technology that Google has been running internally for many years, so it is not a brand new thing," said Tom Kershaw, who is Google's director of product management for the Google Cloud Platform.
Bigtable powers many of Google's core services, including Google Search, Gmail, and Google Analytics.
The service could be used to store sensor data from an Internet-of-things monitoring system. Finance companies could house petabytes of trading data on the service to analyze for emerging trends. Telecommunications companies, digital advertising firms, energy, biomedical, and other data-intensive industries might benefit from the technology as well.
Google Cloud Bigtable is a hosted NoSQL data store. Customers can read and write data using the API (application programming interface) for Apache HBase, which is an open-source implementation of the Bigtable architecture for storing data across multiple servers.
Because Google Cloud Bigtable can be accessed through HBase commands, customers can easily use the service with existing Hadoop software. Hadoop is a popular open source data processing platform for working with extremely large sets of data. Google Cloud Bigtable can also work with other Google cloud services, such as Google BigQuery and Google Cloud Dataflow.
The company claims that Google Cloud Bigtable is speedier than other NoSQL data stores. In one internal benchmark, Bigtable offered a lower latency for reading and writing data than either a generic version of HBase, or the Cassandra NoSQL database.
Google fully manages the service. It handles data replication for backup, and encrypts the data for security. Users can spin up a new Bigtable cluster very quickly. As the data set grows, Google automatically provides the additional storage capacity.
A number of third-party vendors have already rolled Bigtable into their own sets of services. For instance, financial software and services company Sungard has built a financial audit-trail system on Cloud Bigtable that can ingest 2.5 million trade messages per second.
Pricing for Google Cloud Bigtable will be based on a number of factors, including network usage, number of nodes deployed, and amount of storage used. Those interested in testing the service, now in beta, can sign up for a free trial.