After two years in development, Splice Machine Inc. is launching what it calls the world’s only Hadoop-based relational database management system (RDBMS) into general availability with the promise of providing a cheaper and more scalable alternative to the Oracle Corp. databases that have ruled the enterprise for the last few decades. While the idea of running an RDBMS on top of Hadoop may seem like a contradiction in terms, Splice Machine says there are good reasons to do it.

Yahoo! Inc. built Hadoop to fill an internal requirement for a way to cost-effectively ingest large volumes of unstructured data amassed over long periods of time. Relational systems, in contrast, are designed to process fast-moving structured transactions in near real-time. They’re also notoriously expensive, with Oracle Database costing upwards of tens of thousands of dollars per individual processing core depending on the features a customer chooses to enable.

Splice Machine provides what it describes as a full-featured alternative to RDBMS at only a fraction of the cost. The startup can’t boast of spending the last three decades improving its platform like Oracle did, but it’s touting a value proposition that it hopes it attractive enough for that not to matter.

The product is a modified version of the open-source Derby relational database bolted onto HBase, a column store that in turn runs on top of the Hadoop File system and has most of the qualities organizations require for transactional workloads. These include response times fast enough to facilitate interactive analysis, fault-tolerance and a a high level of row-level consistency. The startup’s platform extends that reliability to meet the ACID requirements for transaction processing and layers an extra set of value-added capabilities on top.

The new launch release brings with it a host of new additions, including support for window functions the enable more gradual data analysis and the ability to export information into spreadsheets. For admins, version 1.0 of Splice Machine’s system introduces a new console that provides insight into how fast queries are executed and packs security functionality that makes it possible to authenticate users and limit their access to only the data they require to perform their jobs. Rounding out package are native backup and recovery functions.

Under the hood, Splice Machine has added integration with the HCatalog access management technology for Hadoop, which enables customers to take advantage of some of the most important components in the upstream ecosystem. The roster includes the Hive data warehouse, the Pig abstraction layer for the MapReduce execution engine and the data crunching engine itself along with Spark, an alternative that is described as up to 100 times faster when running in-memory.

The database is already used at 20 organizations, according to Splice Machine, which is twice as many it reported eight months ago after closing its second funding round. To make it easier for organizations to move onto its platform and thus hopefully accelerate adoption, Splice Machine is also launching a new Safe Journey program that offers to smooth the journey with training, a migration toolkit and optional consultancy services.

Photo via Pixabay