VoltDB Reaches Out to Hadoop for Fast Data Analytics
VoltDB has made a name for itself by building a fast in-memory distributed relational database that processes transactions and does some real-time analytics too. With today’s launch of VoltDB version 5, the company is smoothing its integration with Hadoop in hopes of helping customers connect the dots between big data and fast data analytics.
Hadoop is increasingly the repository for organization’s big unstructured data, such as call logs, click streams, and location information. All that stuff goes into the Hadoop Distributed File System (HDFS), where data scientists and analysts can use advanced algorithms and data mining tools to tease patterns out of it and devise new business strategies based upon them.
VoltDB, on the other hand, does analytics in the now. The product caters to clients like banks and telecos that not only need to process transactions quickly—VoltDB says it’s SQL database is ACID compliant–but also do some lightweight analytics upon the fresh data. VoltDB wasn’t designed for big data analytics jobs, such as performing feature extraction upon hundreds of terabytes of data. But it’s increasingly being utilized to do fast data analytics.
The NewSQL database company’s increasing focus on fast data analytics and the Internet of Things has brought it closer to the world of Hadoop. While the Hadoop ecosystem is advancing at its own pace towards in-memory analytics and stream processing through affiliated projects like Apache Spark, Apache Storm, and Kafka, the folks at VoltDB realized that it was time to make its software a first-class citizen when it came to integrating with Hadoop.
To that end, today’s release of VoltDB version 5.0 features a slew of new connectors to Hadoop ecosystem data providers. It introduces new connectors for exporting data to HDFS, to Kafka, RabbitMQ, Avro, and HTTP. It also brings a new data import options, including a Kafka loader, which will be handy for using VoltDB to analyze fast-moving data streams; a Hadoop output format loader; and loaders for JDBC and Vertica user defined extension (UDx) files. (VoltDB was founded by Mike Stonebreaker, the MIT professor who also counts Ingres, Postgres, and Vertica among his credits.)
Bruce Reading, President and CEO of VoltDB, says version 5 will help satisfy demand among developers for creating fast, streaming applications with real-time analytics. “As the popularity and adoption for Hadoop continues to surge, there is an increased need for integration between fast and big data so developers can focus on the applications and not the infrastructure,” he says.
The Bedford, Massachusetts company also introduced a new VoltDB Management Center (VMC) with this release. The VMC is a browser-based console that provides all the monitoring and configuration management needs of system administrators.