Hewlett Packard Enterprise Co. today is simplifying its Ezmeral data analytics portfolio, brushing up the user interface, expanding the selection of open-source tools it offers, integrating object and streaming data sources, and improving interoperability across multiple clouds.

The company says it’s attacking the problem of data silos by offering a complete data and analytics platform that is independent of the underlying infrastructure. The enhanced Ezmeral Data Fabric Software provides a federated view of data across hybrid and multicloud deployments for more straightforward access and governance while fostering self-service use of managed and curated open-source tools.

With expanded support for files, objects, tables and streams, the software suite works across multiple data sources and formats and in both public and private cloud deployments, HPE said. It is now also available for the first time as a software-as-a-service offering with automated policy management that lets users optimize workload placement.

Open source commitment

The Ezmeral Unified Analytics Software suite encompasses a variety of open-source tools delivered as a service (pictured). They include the Apache Airflow workflow management platform, Apache Spark analytics framework, Apache Superset visualization platform, Presto SQL distributed data store and Ray distributed computing framework.

For training and deployment of artificial intelligence applications, Ezmeral will support the Feast feature store for machine learning, the Kubeflow machine learning platform on Kubernetes, and the MLFlow machine learning management platform. New connectors and data sources are also being introduced for Snowflake Inc.’s Data Cloud, MySQL, Databricks Inc.’s DeltaLake and data warehouses from Teradata Corp. and Oracle Corp.

HPE said the offering is hybrid by design and can be deployed across edge, colocation, on-premises and public cloud environments. The location-agnostic deployment also reduces data ingress and egress fees.

‘Momentous’

“This is momentous for us,” said Mohan Rajagopalan, vice president and general manager of HPE Ezmeral Software. “We are coming to the market with a portfolio that addresses both data and the next generation of tools. In the past, we had almost a dozen offerings from Ezmeral. That has now been simplified to two.”

Rajagopalan said the announcement underscores HP’s commitment to supporting open-source projects with managed services that eliminate the need for customers to apply patches and upgrades while enabling them to build applications that work across multiple platforms.

“We don’t want to create a fork,” he said. “We want to encourage true open source. We take care of updates and patching but you have the freedom to integrate as you want.”

Among the supported platforms is GreenLake, HPE’s line of hardware and software products that enterprises can buy on a cloud-like pay-as-you-go consumption basis.

“Many of our customers are still on-prem in air-gapped environments,” Rajagopalan said. “We’ve integrated with GreenLake to deliver to them using a SaaS delivery mode.”

He said the expanded support for objects and streams is part of a bigger long-term strategy to work with data in any format. The company plans to add support for graphs and tables in future versions of Ezmeral.

Image: HPE