Research house finds buying the full stack from one vendor is a good idea, with some provisos – most especially interoperability.
International Data Corporation (IDC) has published two reports. The first defines an interoperable framework for infrastructure stacks to deploy AI applications, referred to as the AI Plane (AIP). The second discusses vendor-specific implementations of AI infrastructure stacks.
Overall the research house is in favour of these integrated stacks, believing they bring benefits quickly, but urges thorough investigation before making procurement choices and above all, advises companies to consider the stacks’ interoperability.
As building AI capabilities becomes increasingly urgent, IDC sees that businesses are confused about the process of building their own AI infrastructure stack.
At the same time, more AI server, storage, and processor vendors are developing AI stacks that consist of abstraction layers, orchestration layers, AI development layers, and data science layers that are intended to seamlessly operate together.
The stacks typically combine open source software, proprietary software, and “non-monetized” commercial software layers, such as Compute Unified Device Architecture (CUDA) – a parallel computing platform and application programming interface (API) model created by Nvidia.
They are intended to help customers’ IT infrastructure teams, developers, and data scientists collaborate on a complete stack without having to build it themselves.
IDC believes that AI infrastructure stacks provide a clear advantage to customers and that their variety is, while confusing, not a disadvantage.
However, the research house does not expect vendors to collaboratively develop a “standard” AI infrastructure stack – this would defeat the advantage of such broad choice for customers.
By offering an AIP framework, IDC hopes to provide a guide for IT vendors, encouraging them to improve the versatility of their stack, thereby increasing AI’s ubiquitous adoption.
IDC recommends that technology buyers thoroughly investigate the entire AI stack that server vendors offer and to explore options beyond their usual hardware supplier.
IT benefits to keep in mind when assessing reference stacks include reduced costs, data and application availability, effective infrastructure consolidation and, where possible, a single interoperable application delivery platform.
IDC also recommends technology vendors to focus on interoperability among AI infrastructure stacks.
“Businesses are benefiting tremendously from the AI infrastructure software stacks that server, processor, and co-processor vendors are making available, several of which we are highlighting in these reports,” said Peter Rutten, Research Director, Infrastructure Systems, Platforms and Technologies at IDC.
“But buyers should be aware of complexity and a lack of interoperability with these stacks.”
Sriram Subramanian, Research Director, Infrastructure Systems, Platforms and Technologies at IDC, continued, “Infrastructure requirements for AI workloads can be viewed as a function of scale, portability, and time.
“With so many choices and options available, end users are often perplexed about the right infrastructure stack. AIP provides a simple framework to select the right infrastructure stack, with accommodations to considerations on cost, flexibility, and infrastructure utilization.”
The IDC report, The AI Plane: An Interoperable Framework for Artificial Intelligence Infrastructure Stacks (IDC #US46283420), introduces AI Plane (AIP) – an interoperable framework to select the right infrastructure stack to power AI workloads. The report introduces two specific implementations of AIP: Open AI Plane and as-a-Service AI Plane. IDC recommends that enterprises leverage the AIP framework when selecting an appropriate infrastructure stack to power AI workloads.
The IDC report, AI Infrastructure Stack Review H1 2020: The Rapid and Varied Rise of the AI Stack (IDC #US46291620), assesses some of the AI infrastructure stacks that are available on the market, including Intel, AMD, NVIDIA, Cisco, Huawei and HPE.