In September, HPE announced the availability of HPE Private Cloud AI, a solution HPE jointly developed with NVIDIA to provide an onramp to building and launching AI applications in minimum time. HPE describes it as a “turnkey, cloud-based experience.”
“There are seven steps along the machine learning operations workflow,” said Alex Ollman, product manager for HPE, while explaining the role of private cloud infrastructures in enterprise AI strategies at the recent AI Data Infrastructure Field Day, an event hosted by Tech Field Day, a Futurum Group company.
These workflows encompass gathering, preparing, choosing, building, training, validating and deploying.
“It requires appreciating all of the underlying infrastructure to perform any one of them,” he stressed.
Born out of the NVIDIA AI Computing by HPE initiative, Private Cloud AI is a solution architected for personas like data scientists, data engineers, ML engineers, AI engineers and app developers who work at the forefront of AI application development in enterprises, and could use tools that save time and energy required for secondary tasks.
“Private Cloud AI is an infrastructure stack that is being sold as an appliance to be able to perform data engineering, data science and machine learning in the age of generative AI in the same way that a microwave is an appliance abstracting a number of different things required to make and heat up food,” explained Ollman.
Private Cloud AI performs abstractions up and down the layers of software, hardware, networking and compute. The solution is enabled by HPE GreenLake Cloud and offers a central platform-based management portal to automate, orchestrate and mange users and data in the hybrid ecosystem.
The solution includes best-of-breed hardware constituting AI servers from HPE, and NVIDIA GPUs, and “accelerators” as HPE calls them, such as software and services from both vendors and other third-parties, a library of pre-integrated AI models, a GenAI virtual assistant, tools and frameworks, and full-stack support for all HPE and NVIDIA hardware and software solutions.
Private Cloud AI reflects HPE’s ambition to grow out of the contours of an infrastructure company and provide a complete solution covering the hardware, the software and the expertise.
“Through delivering a holistic solution, we wanted to start moving towards solution selling…We saw that people are going to start getting value out of public cloud vendors, but we knew that as time went on, that abstraction of infrastructure will become not only a cost burden, but a burden around data sovereignty and understanding and owning that end-to-end lifecycle of infrastructure.”
Ollman referred to three power users for HPE Private Cloud AI – cloud administrators, AI administrators and AI developers. For cloud administrators, he said, Private Cloud AI will provide the ability to spin up and deliver infrastructure with one touch. AI administrators will find capabilities like quick onboarding of users onto the infrastructure and centralized access management, very handy.
AI developers, the brains behind the development of AI applications, will not have to deal with infrastructure-level complexity that keeps them from being optimally productive because all of it is masked with Private Cloud AI.
“It isn’t the data scientist’s job to have to spin up multiple nodes and create an orchestrated mesh of Spark workers to a Spark controller,” said Ollman. “They just want to run a query on a database.”
According to HPE, one of the solution’s greatest features is that it allows administrators to manage users, assign roles and regulate accesses all from a single plane.
“You can delegate individual users and whole teams not just access to the infrastructure, but what they’re allowed to do on that infrastructure.”
Administrators can tightly control and set limitations on what the users can access and control without logging in and authenticating every user.
“Anyone who has stored an AWS private key in a notepad on their desktop will know what I’m talking about.”
Often confused with the PC vendor, Hewlett-Packard, HPE (Hewlett Packard Enterprise) is an independent company that was founded after splitting off from Hewlett-Packard in 2015.
“The Private Cloud AI offering is the cultivation of that vision that HPE started when it first branched off from Hewlett-Packard all those years ago,” said Ollman.
Watch HPE’s presentations from the AI Data Infrastructure Field Day event for a deep dive of Private Cloud AI.
