Nutanix Inc. today announced a broad set of enhancements to its core infrastructure software and its Project Beacon portfolio of data-centric platform-as-a-service offerings that can run across multiple cloud and on-premises platforms.

The company is aiming to cater to the 80% of organizations that use multiple public cloud platforms or plan to in the future. It’s also looking to steal a share of the 80% of VMware Inc. customers that Forrester Research Inc. expects will abandon the vSphere hypervisor this year after Broadcom Inc.‘s acquisition of the virtualization giant.

Cloud-agnostic Kubernetes

The Nutanix Kubernetes Platform provides a Cloud Native Computing Foundation-compliant stack for securely managing the Kubernetes software container orchestrator across on-premises, hybrid and multicloud environments. Customers can manage multiple clusters running containers on Nutanix on-premises and in the public cloud from a single pane of glass. That includes clusters running in non-Nutanix environments, including the most popular public cloud Kubernetes services, and disconnected “air-gapped” configurations.

“For the first time, we’re not running our stack on a hypervisor either delivered or managed by us,” said Lee Caswell, senior vice president of product and solutions marketing. “It’s an opportunity to run in a containerized manner on infrastructure delivered by someone else.”

NKP integrates with Nutanix’s portfolio of data services, including scale-out block, file, and object storage and databases-as-a-service. It provides high levels of automation and artificial intelligence-powered analysts to reduce complexity. Because it’s built on CNCF-conformant Kubernetes, there is no cloud lock-in, the company said.

The offering comes in three tiers. The Starter tier is included in Nutanix Cloud Infrastructure and effectively replaces the existing Nutanix Kubernetes Engine. NKP Pro adds a suite of cloud-native projects for individual clusters, including built-in Nutanix Data services. NKP Ultimate is for managing large fleets of clusters on-premises and in the public cloud.

“While the majority of cloud-native applications are born in a cloud, very few are actually multicloud and built to  be moved easily from one cloud to another, or on-premises,” wrote Thomas Cornely, senior vice president of product management at Nutanix, “They also lack consistent data services to store, protect and process data across clouds.”

Storage services are also typically specific to the host provider’s cloud platform, requiring additional work by Kubernetes platform engineering teams to deliver consistent services securely and manageably, he wrote.

AOS containerized

As part of Project Beacon, Nutanix is also containerizing its Cloud-Native AOS operating system, with the first use case being Nutanix Data Services. This allows cloud-native applications to remain available during failures in availability zones, regions, and even full cloud services. Cloud-Native AOS will support automated replication of all container data across availability zones and regions. A future version of Nutanix Files running natively on the public cloud will provide enterprise-grade file services.

Nutanix is also adding new options to its AHV hypervisor that let users modernize infrastructure while reusing existing server hardware. The company said it’s working with Cisco Systems Inc. to certify Cisco’s Unified Computing System for AHV. Nutanix plans to expand AHV compatibility across a broader set of server configurations and will support migrating many of the most popular VMware vSAN ReadyNode configurations running on existing hardware.

AHV’s Secure Snapshot capability is being strengthened with a new multiparty approval control for privileged operations such as snapshot changes. AHV will also get enhanced multisite disaster recovery features for recovery from simultaneous site failures. Automatic Cluster Selection is intended to simplify self-service by intelligently placing newly created virtual machines across a set of clusters while balancing resource utilization automatically.

“Clusters can have different characteristics for mixed nodes,” Caswell said. “Placing clusters was manual in the past. Now you tell us the characteristics, and we’ll direct you where to place it in the most optimal manner.”

GPT partnerships

Finally, the whimsically named GPT-in-a-Box toolset for developing generative pre-trained transformer applications Nutanix introduced last year is getting integrations with Nvidia Corp.’s NIM inference microservices and Hugging Face Inc.’s large language model library.

GPT-in-a-Box is described as a full-stack enterprise AI development environment that integrates with Nutanix Objects and Files. GPT-in-a-Box 2.0 allows developers to easily access validated LLMs from Hugging Face. It sports a unified user interface for foundation model management, application program interface endpoint creation, end-user access key management and integration with Nvidia Tensor Core graphics processing units.

For more ambitious AI developers, Nutanix enhanced its unstructured data platform with increased performance and density. Nutanix Unified Storage now supports a new non-volatile memory express-based platform with more than 550 terabytes of storage and up to to 10 gigabit-per-second sequential read throughput from a single node. Nutanix will also support Nvidia GPUDirect Storage, a family of technologies that establishes a direct data path between GPU memory and local or remote storage devices.

The company also announced an AI Partner Program that provides simplified access to an ecosystem of generative AI partners.

Photo: Flickr CC