Kyvos serves up Hadoop on cubes
- 01 July, 2015 08:12
New software from Kyvos allows users to build data cubes from Hadoop data
What good is big data if you don't have the proper tools to make sense of it? A US-based startup, Kyvos Insights, can transform terabytes of information stored on Hadoop clusters into easy-to-parse data cubes, the format business analysts prefer.
"We focus on making big data interactive and usable to the business user," said Ajay Anand, Kyvos's vice president of products. "We can do online analytical processing [OLAP] at massive scale."
The Kyvos software can work with any major Hadoop distribution, including those from Cloudera, Hortonworks and MapR, and the stock distribution from Apache Software. Users can parse the results using Kyvos's own software or third-party programs such as Microsoft Excel or Tableau's data visualization software.
The aim of the software is to format the unstructured data captured by Hadoop into the familiar OLAP patterns that form the basis of much business intelligence software. In particular, the software can merge data from different sources into a single data cube.
A data cube is a multidimensional array of values, often derived from two or more database tables. It's useful for comparing data across different silos, Anand said. For example, a company may want to look at sales figures from different regions to spot trends that aren't evident from examining any one area. All that information can be merged into a single data cube for easy inspection.
The Kyvos software is one of a number of Hadoop-based offerings that tie together the worlds of big data and traditional business intelligence. Apache itself offers the open-source Hive, which allows users to create a Hadoop-based data warehouse. Kyvos is unique in that users don't have to buy additional hardware to set up a stand-alone data warehouse or data mart, Anand said. Instead, Kyvos can use the existing Hadoop cluster to store the data.
One early customer has been Entravision Communications, based in Latin America, which needed to pull together different sources of information to evaluate new business opportunities. The company used the software to merge profiles of 15 million of its customers, work that otherwise would have been too unwieldy to undertake using traditional business intelligence tools.
Analyzing credit-card transactions and shopping and travel patterns would be other workloads suitable for Kyvos.
A typical implementation will cost about US$15,000 a year and can be run on premises or in the cloud.