carloscastilla - Fotolia

Kyligence builds out data cloud for OLAP and big data

Kyligence is advancing the Apache Kylin project with a cloud-native offering that can help organizations more efficiently execute and manage data queries against large data sets.

Open source-based startup Kyligence is advancing the capabilities of its Online Analytical Processing data warehouse with the general availability of the Kyligence Cloud 4 update on Jan. 21.

At the core of the Kyligence platform is the open source Apache Kylin project, which was started at eBay in 2014, providing an OLAP data warehouse capability for big data Hadoop workloads.

A key part of Kylin is the ability to enable OLAP cubes, which provides a database structure that can visualize and compute analytical data in an optimized approach that accelerates queries. The startup, based in San Jose, Calif., builds on top of the open source Kylin project providing commercial support and enterprise features for automation and machine learning.

With the Kyligence Cloud 4, the vendor is introducing multiple enhanced capabilities, including a unified semantic service that helps centralize data from multiple sources. The new release also provides performance enhancements designed to accelerate data queries on large data sets.

Accelerating OLAP cubes with Kyligence

Mike Leone, senior analyst at Enterprise Strategy Group (ESG), said that as organizations rely more heavily on data, more people want access to it, which can be a challenge for OLAP cubes.

"For a traditional OLAP approach to solve the needs of an organization that's exploding with data, it could take thousands of cubes, all of which require management and maintenance, and that's before the budgetary nightmare," Leone said. "Kyligence is focused on solving the complexities that more and more organizations are suffering from in association with greater scale of data, more urgency for better, faster insights and taking back control of unruly deployments."

Leone added that with the new Kyligence Cloud 4, the vendor is delivering a cloud-native architecture that can be rapidly deployed and includes a unified semantic layer that benefits from the intelligence of an AI engine to auto-model and self-tune cubes.

Screenshot of Kyligence Cloud architecture
The Kyligence Cloud architecture enables users to take different data source types and enable them for accelerated online analytical processing (OLAP) queries.

How Kyligence Cloud 4 accelerates OLAP

Li Kang, head of North America operations at Kyligence, explained that the vendor's platform can communicate with any data source to read data. Kang emphasized that Kyligence Cloud 4 does not move data from the source, but rather executes what he referred to as "precomputation" of results.

Precomputation is the foundational computer science concept behind OLAP cubes, providing an initial computation of a data set, such that future queries can be expedited.

Kyligence is focused on solving the complexities that more and more organizations are suffering from in association with greater scale of data, more urgency for better, faster insights and taking back control of unruly deployments.
Mike LeoneSenior analyst, Enterprise Strategy Group

Kang explained that the precomputation results are stored in distributed OLAP cubes inside of a Kyligence Cloud deployment. The precomputation data is all stored by Kyligence in the open source Apache Parquet data format.

Kang noted that Kyligence also uses the Apache Spark query engine to build the OLAP cubes from the original data sources. The front end of the data is exposed in Kyligence via a unified semantic layer, so that an organization can use its own choice of business intelligence or machine learning algorithms.

"This gives enterprise users a single view of the data, as well as the semantic layer to connect different front-end tools or products," Kang said.

Another core element of the Kyligence Cloud 4 release is an AI engine that also helps accelerate OLAP queries. The AI engine is constantly learning from queries and data profiles to help improve and optimize the OLAP cube.

"Once you start using the product, every time you're sending a new query to the product, the engine will learn from this new query and will try to further optimize the cube," Kang said. "That kind of optimization is done behind the scenes, so that we can keep the system at the best performance and also best user experience."

Enterprise Strategy Group (ESG) is a division of TechTarget.

Dig Deeper on Data warehousing

Business Analytics
Content Management