With the new services for HPE GreenLake, Hewlett Packard Enterprise (HPE) aims to combine the advantage of a local data center with the scalability of a Cloud connect. For this purpose, the group offers supercomputers that are installed locally at the customer's site, but are still billed as an as-a-service solution like a cloud only according to the actual workload. HPE continues to take over the operation.
The new billing model should enable companies to use the power of a supercomputer without large initial investments. Three service packages with different performance are offered, which customers can order via a self-service portal. According to HPE, the delivery time for the supercomputers is only 14 days.
According to a study by Hyperion Research, the most common concerns before purchasing a supercomputer are the high one-off acquisition costs, in addition to the operating costs for power and cooling and the specialist personnel required in the field of high-performance computing. HPE's new sales model therefore removes an often existing obstacle and could thereby encourage more companies to use their "own" supercomputer.
HPE Apollo systems as the basis for supercomputers
The new supercomputer-as-a-service solution will initially be offered on the basis of HPE Apollo systems. Included are software solutions such as workload management, containers, orchestration, cluster management and monitoring.
In future, HPE's entire supercomputer portfolio is to be offered as an as-a-service solution. In principle, therefore, supercomputers are also conceivable which could enhance the performance of the Cerebras CS-1 reach. In addition, the offer is to be extended to Cray-based computing, software, storage and network solutions.
Provision via partners
The supercomputers are provided in cooperation with local partners, including system houses and colocation facilities. This makes it possible for customers who do not have the opportunity to operate a supercomputer locally to set it up in the computer center of an HPE partner. The computing power can then be used remotely via HPE GreenLake.
In addition, companies such as Activeeon, Ansys, Core Scientific and TheUberCloud are expected to join HPE in providing artificial intelligence (AI), automation, computer aided design (CAD) and aided engineering (CAE) solutions to end customers. This alignment underscores that HPE, as well as. Amazon Web Services (AWS) is attempting to increasingly address industrial customers in addition to traditional software companies.
Administration and cost overview via software
The central administration of the supercomputers is done by the tool HPE GreenLake Central. The HPE Self-Service Dashboard for the operation of the high-performance cluster. HPE says that configuration changes can be made without interrupting work. Using the HPE Consumption Analytics tool, customers can also report their usage and the Costs of their HPE GreenLake supercomputer.
HPE's first customer is Zenseact, which is researching solutions for autonomous driving in China and Sweden. The supercomputer is used there to analyze several hundred petabytes of data. Datagenerated from a network of test vehicles. Additional customers are expected to receive initial HPE GreenLake Cloud Services in spring 2021.
