HPE Ezmeral Runtime Enterprise 4‑year 24x7 E‑LTU

Decoupling of Platform and the Kubernetes-related Components

HPE Ezmeral Runtime Enterprise will decouple the upgrade of platform and K8s-related components, reduce downtime, and give the administrator an ability to upgrade K8s-related components without performing a complete platform upgrade.

The Kubernetes-related components will be delivered in Kubernetes Bundles, which will package the software to support newer Kubernetes versions and updated add-ons. See Kubernetes Bundle for more details.

No changes were made to the upgrade procedure of an existing Kubernetes cluster and if the Kubernetes version update requires changes in the HPE Ezmeral Runtime Enterprise platform, the administrator can perform the complete HPE Ezmeral Runtime Enterprise upgrade.

Support for SLES 15 SP3, Newer K8s Versions, and External HPE Ezmeral Data Fabric on Bare Metal 7.0

The HPE Ezmeral Runtime Enterprise 5.5.0 release adds support for SLES 15 SP3 OS, K8s versions 1.21.14-hpe1, 1.22.12-hpe1, 1.23.9-hpe1, and open-source components such as Istio® (1.13.5). The K8s versions are based on the HPE with K8s distro, hence the “-hpe1” suffix.

It also adds support for external Ezmeral Data Fabric on Bare Metal 7.0 and flexibility around data fabric registration by allowing multiple Ezmeral Runtime Enterprise instances to register with the same external HPE Ezmeral Data Fabric on Bare Metal cluster as tenant storage.

Support for Apache Spark 3.2, 3.3 and GPU Acceleration for Spark Workloads

HPE Ezmeral Runtime Enterprise includes HPE Ezmeral Spark version 3.2 compatible with HPE Ezmeral Data Fabric 7.0, along with user authentication and governed data access to the data fabric. In addition, the Spark Operator has been enhanced to support open-source Apache Spark versions 2.x and 3.x.

Spark workloads can use GPU acceleration through built-in support for the NVIDIA® RAPIDS plugin.

Enhanced ML User Experience and Secured Model Management in a Multi-tenant Environment

Improves data scientist productivity to both build and operationalize ML models quickly.

Streamline access to data sources, notebooks, experiment tracking and model registry from a single, intuitive UI.

Built-in support to use MLflow integrated with Kubeflow for experiment tracking and model management.

Shared MLflow model registry and experiment tracker across tenants; end users are authenticated and provided controlled access to model and model metadata.

Includes support for Kubeflow 1.6