HPE Ezmeral Unified Analytics Software open and extensible platform is designed to simplify building, deploying, managing, and monitoring analytics and AI models.
Artificial intelligence (AI) is the most data and computationally intensive workload of our time and requires an evolution of IT architectures to utilize an enterprise’s proprietary data to effectively train, tune, deploy, and monitor AI models. Today, every modern enterprise has AI-native as one of their top priorities — be it to delight customers, drive business growth or deliver competitive differentiation.
I have spent many months speaking with global customers across healthcare, finance, banking, gaming, insurance and government and they are all struggling to build a modern data analytics & AI-native strategy. The struggles are a result of hurdles — both business and technical — that need to be overcome for any successful AI-native strategy.
Comments from customers have been around these themes:
The gap between Gen AI hype and reality is wide. While Generative AI is top of mind for CXOs, there is a still a significant gap or a challenge to build and operationalize AI systems. Too many bespoke tools, lack of standards, inconsistent experience are some of the impediments to success. As a result, enterprises are still not seeing meaningful ROI from their investments.
Today’s data is highly distributed across on-premises and multiple cloud environments making it difficult to deliver high-integrity data pipelines to analytics and AI teams. Traditional analytic approaches use multiple point solutions for each environment increasing complexity that slows down unification and the insights used by data consumers.
High public cloud costs are causing organizations to shift away from a cloud-first to a hybrid by design strategy for analytics and AI-native initiatives. Organizations are weary of the sticker shock associated with public cloud resources as well as lock-in to a proprietary vendor’s IP. Organizations are looking for a unified approach to analytics and AI-native initiatives across hybrid environments with full cost transparency.
Unable to productize open source innovations: In recent years, robust open source tools and frameworks have become a viable alternative to specialized commercial software. While open source innovation is attractive, the struggle to secure, scale, upgrade, and maintain these tools has become a new pain point leaving customers with a difficult choice to being locked into a solution or face the cost of securing and maintaining open source tools and frameworks.
Data and AI for any modern hybrid enterprise
HPE Ezmeral Unified Analytics Software brings together a fully managed open source ecosystem with security and automation enabling organizations to develop, deploy, manage, and monitor data analytical and AI-native workloads across hybrid and multi-cloud environments.
The goal of this solution is to overcome the challenges of working across hybrid environments. In short, this solution provides:
- Increased productivity for data analytics and AI teams through self-service access to unified data, self-service to the open source tooling they prefer, and comprehensive security.
- HPE validated and managed open source ecosystem, segmented by persona, that is available with a single click. Quarterly updates from HPE ensure that data-first and AI-native teams are using the freshest open source innovation.
- Self-service access to popular business and analytic data sources from a built-in catalog.
- Extensible platform that enables analytics and AI teams to import custom open source apps and third party ISV solutions along with the added benefits of enterprise security, observability and scaling.
The latest release of HPE Ezmeral Unified Analytics Software enhances the solution’s capabilities with the following innovations:
Expanded hybrid cloud deployment targets: HPE Ezmeral Unified Analytics Software can now be installed on Microsoft Azure and Google Cloud Platform (GCP). When combined with existing Amazon Web Services (AWS) support, allows organization to drive faster time to insights by bringing analytics closer to your data.
HPE Machine Learning Development Environment (MLDE): HPE Machine Learning Development Environment and HPE Ezmeral Unified Analytics now work together allowing organizations to build, train, and operationalize ML, DL, and LLM models on a single platform.
Advanced GPU management: The latest release enables organizations to schedule their scarce GPU resources by reclaiming idle or unused GPUs from one application and assign it to higher priority workloads without manual intervention. Usage of GPUs at the workload level can be monitored from the intuitive console making it easier than ever to align workloads and resources to your business goals.
Model monitoring: This release includes an early preview of model monitoring allowing organizations to continuously monitor model quality machine learning model quality to safeguard reliability across hybrid environments. By leveraging open-source Whylogs libraries, organizations can quickly identify issues and take corrective measures, such as model retraining.
Elevated security capabilities: Security is a non-negotiable basic for every organization which is why HPE Ezmeral Unified Analytics Software comes with zero-trust security built-in. In the current release, we’ve added several capabilities to provide multi-user resource isolation and a secure sandbox environment across open software tools such as Airflow, Spark, MLFlow, KubeFlow, Superset and others. Every user can see and act on their own resources be it data pipelines, ML experiments, Spark jobs or visualization charts.
Second, we’ve added uniform security across HPE Ezmeral Unified Analytics Software and HPE Ezmeral Data Fabric Software to ensure secure access to data across the platform.
Third, for authorized users authenticated into HPE Ezmeral Unified Analytics Software, we provide seamless conversion of user credentials to access objects in S3 buckets without users supplying access and secret key credentials.
Expanding data foundations: To enable self-service data access, HPE Ezmeral Unified Analytics Software provides connectors to popular data sources including HPE Ezmeral Data Fabric Software, on-premises and cloud data lakes, lakehouses and data warehouses, OLTP databases. In the latest release, we have added Iceberg connectors to allow access to data in popular Apache Iceberg open table format. This is in addition to Delta Lake table format that was already supported. With a built-in data catalog, users can select datasets that suit their needs, access them across any tool or app within HPE Ezmeral Unified Analytics Software.
Upgrades made easy: All open source tooling and frameworks can now be upgraded either with the push of a button or through scheduled upgrades. The eligible upgrades are pushed to HPE Ezmeral Unified Analytics Software instances via SaaS control plane. Upgrades happen in the background and admins do not have to intervene with any further action. Complete support matrix of various OSS versions can be found here.
In summary, HPE Ezmeral Unified Analytics Software provides a foundation for data and AI professionals to develop, deploy, manage, and monitor workloads across hybrid environments allowing your business to innovate faster, lower costs, and remove.
Ready to get started?
To learn more, please visit https://developer.ezmeralsoftware.hpe.com/
If you are attending HPE Discover, Barcelona, consider attending these sessions:
Building the data foundations for advanced analytics and AI, Session, Catalog #6355
Future proof hybrid SaaS foundations for data and AI, Catalog #6346
Srikanth is a customer-centric, data-focused product leader with over 20 years’ experience in building innovative hybrid-cloud enterprise software products that tackle real-world data management & AI challenges. Connect with him on LinkedIn.
Ezmeral Experts
Hewlett Packard Enterprise
twitter.com/HPE_Ezmeral
linkedin.com/showcase/hpe-ezmeral
hpe.com/Ezmeral
