Clouds for AI and HPC

While HPC has traditionally been the purview of academic research and national laboratories, it is now going mainstream as a way to derive value out of growing stores of data.

cloud network

The unmitigated growth of data, along with the need to glean insights from that data, is driving the convergence of artificial intelligence (AI), high performance computing (HPC) and cloud, highlighting the advantages of a hybrid, multi-cloud IT operating model. Organizations looking to adopt data science practices using AI and HPC stand to benefit greatly from multi-cloud choice, efficiency, and scalability.

While HPC has traditionally been the purview of academic research and national laboratories, it is now going mainstream as organizations look to harness HPC, along with AI, to derive value out of growing stores of data. And this is forcing IT decision makers to consider the cloud to improve infrastructure efficiency and scalability, according to a recent Forrester study.

The success of AI initiatives requires a modern IT infrastructure, one that scales across organizations, including everything from servers, storage, and networking through to data warehouses and cloud. Improved infrastructure utilization, scalability, faster data analysis and the use of myriad data services are key drivers of increasing cloud usage.

In online surveys with over 700 IT decision makers, Forrester found that many are deploying AI and HPC on cloud and this trend will continue. Other key findings include:

  • Hybrid cloud is the new reality. “Organizations are making increasingly more use of cloud — both public and private — especially when it comes to HPC and AI workloads. But this doesn’t mean all deployments are moving off-premises. In reality, respondents say their organizations expect cloud use to increase moving forward, but most are still working with environments that combine both cloud and non-cloud infrastructure.”
  • Organizations must contend with security, latency, and application architecture issues while managing costs. “The majority of respondents report struggling with security issues, network delays, and necessary app rework that can derail projects and lead to increased costs. Though managing costs is a universal goal, firms need to put a heavier focus on cost optimization if they want to be successful.”
  • Hybrid cloud can be a cure for what ails organizations. “Cloud migration will always be difficult. However, respondents in our study find that deploying HPC and AI workloads specifically in a hybrid environment results in business benefits that directly solve for some of their biggest challenges, most notably: improved performance, lower costs, and better security.”
  • Mature cloud use can transform your business. “Respondents say their firms are already reporting improvements to their bottom lines from their move to cloud. What’s even more encouraging? Some mature organizations have instituted formalized processes, compiled sufficient resources, and maintain consistent management to help them use cloud as a way to fundamentally transform their business. As a result, these firms enjoy even larger gains from their efforts.”

Forrester has stated that multi-cloud is here to stay, and most survey respondents agree. The fact is that many organizations today pursue a hybrid cloud strategy with a mix of public and private clouds based on application and business requirements. And these hybrid cloud environments are proving quite popular for AI and HPC.

Some of the key benefits noted in this approach include:

  • “Improved IT infrastructure management and flexibility”
  • “Ability of IT to focus on strategic projects rather than on managing the datacenter”
  • “Improved application and/or infrastructure performance”
  • “Access to temporary or peak capacity to meet demand surges”
  • “Better IT cost management”
  • “Greater agility to respond more quickly and capably to workload demands”
  • “Lower compute and storage costs”
  • “Improved security and compliance.”

These findings, and more, lead to the conclusion that when all things are considered, the convergence of AI, HPC and cloud will prove to be a boon for organizations as they pursue data science initiatives. However, it will require a systematic approach, across an organization, to build out the modern IT infrastructure needed for success.

To Learn More

Explore more AI and HPC solutions from Dell Technologies as well as the AI and HPC offerings from Intel.


Copyright © 2021 IDG Communications, Inc.