AtScale Inc. today updated its business intelligence abstraction platform with support for data lakes of any size and simpler migration of analytics workloads across business intelligence tools.

The company said AtScale 6.5 enables users of BI tools to share queries freely with each other, regardless of which front end they prefer.

AtScale’s technology provides an abstraction layer for a Hadoop cluster or other back-end data store so that it can be accessed by business intelligence applications without the need for extensive data extract/transform/load procedures. The new release expands the semantic layer to cover data lakes, which are repositories of structured and unstructured data that are typically used for business analysis. This permits queries to be run against datasets of virtually any size with processing on the server rather than inside the desktop BI tool.

The company said it’s fundamentally eliminating the need for data marts, which are small databases that contain data that has been extracted from a data lake or warehouse and cleansed for end-user access. Use of the semantic layer enables all users, regardless of the platform they prefer, to use one set of universal semantics to query any data. A query written in Microsoft Corp.’s PowerBI can thus be run unchanged by someone using Tableau Software Inc. visualization engine, the company said.

Many tools, one query

“We’ve added what’s essentially a business analysis optimization layer that reads all reports and virtualizes them into a single semantic layer,” said Dave Mariani, AtScale’s co-founder and chief executive. Migrating BI reports and dashboards to consolidated and consistent models based on a single set of query parameters promote business logic reuse and enable business logic to be shared among all BI tools in the organization.

AtScale said its caching system provides fast and scalable percentile estimation capabilities on extreme data sizes so that enterprises no longer need to subset their data or wait for it to be loaded into a data mart. “We’re doing the calculations on the raw data on the underlying data platform,” Mariani said. “Because we operate on the data itself, and not on an extract, we’re able to operate at a much larger scale.” The system works with both on-premises and cloud data stores.

Also added with this release is a set of profiling features that enable different groups within an organization to have unique and secured views of a common data queue. This eliminates the need for IT organizations to build, materialize, store and maintain different views of the same data for each group.

“I can hide or show levels of a hierarchy based upon what group privileges I’ve assigned with my directory,” Mariani said. “That means I can build one master cube for a subject area and lock down elements that I don’t want to be visible to the entire group.” Those permissions are inherited at the semantic layer level so that they remain consistent regardless of the BI tool being used, he said.

The company didn’t disclose pricing information. Founded in 2013, AtScale has raised $45 million in funding, including a $25 million round last October.

Matthew Baird, co-founder and chief technology officer of AtScale, talked about abstraction layers and the evolution of data lakes today with Lisa Martin and George Gilbert, hosts of SiliconANGLE’s livestreaming studio theCUBE, at the Big Data SV 2018 conference in San Jose, California. (Disclosure: AtScale sponsored this segment of theCUBE. Sponsors have no influence on content on theCUBE or SiliconANGLE):

Image: Unsplash