[This 2 part blog series will explore deployment of full stack HCI infrastructure with acceleration technologies validated through a set of reference architectures built using VMware Cloud Foundation, VMware Tanzu and Intel® Optane™ persistent memory, second generation Intel Xeon™ scalable processors along with Intel Deep Learning Boost technologies. Thanks to the Intel Data Center and VMware OCTO teams for collaborating on this exciting program.]
To create strategic differentiation through the pandemic, many organizations are balancing ongoing cost reduction efforts with digital transformation initiatives, fueling the need to increase innovation and accelerate product development to create competitive advantage. At the center of this wave are Machine Learning and Artificial Intelligence (ML/AI) technologies that have become ubiquitous across every industry sector. This is driving tremendous growth, with IDC forecasting that global AI technology spending will increase from $50.1B (2020) to more than $110B in 2024[1]. As organizations target the initiatives that will increase competitive advantage, modern business leaders understand that reinvesting cost savings to drive digital business initiatives are the best way to realize ROI across a broad range of use cases.
Delivering on the Promise of Hybrid Cloud
Now, as hybrid cloud and distributed compute models become mainstream, infrastructure and application teams can deploy advanced AI and machine learning workloads on-premises to the edge and to public cloud, building upon an agile infrastructure platform that can meet the needs of these dynamic, high performance application environments. Working very closely with VMware, Intel has built a reference architecture delivering a Hybrid Cloud Analytics Solution that transforms the infrastructure to support higher levels of analytics performance, improved efficiency resulting in drastic improvements in Machine Learning benchmarks and model training, in addition to performant processing of mainstream enterprise databases.
Figure 1: ML/AI Hybrid Cloud Reference Architecture with VMware Cloud Foundation[2]
This performance is highlighted across real world use cases to demonstrate the value of the machine learning platform, including data warehousing, advanced analytics, machine learning training and inferencing. One such example modeled the COVID-19 data at a County Level and that model is available here and shown below.
Figure 2: Data Set COVID-19 Cases by County
The Path Forward
Growth in machine learning, AI and advanced analytics is being driven by a completely new generation of applications that are placing extreme demands on compute infrastructure, utilizing advanced silicon and hardware acceleration to increase performance with the aim of offloading cycles from the CPU. This is also creating fundamental changes to the infrastructure architecture in order to support this next generation of applications. Project Pacific, announced at VMworld 2019 and delivered this year as vSphere with Tanzu is a strong example of how the next generation of container-based cloud-native applications have resulted in architectural changes in core infrastructure solutions. Project Monterey, announced as technology preview at VMworld 2020, will take these architectural changes to the next level by fully leveraging Smart NIC technologies from leading partners to offload the analytic workload requirements from the CPU.
VMware Cloud Foundation Accelerated by Intel Provides an Agile AI Platform
What do all of these advanced analytic and application workloads have in common? They all utilize VMware Cloud Foundation (VCF) as the underlying cloud platform to automate delivery of these advanced services taking advantage of the full stack HCI architecture, purpose built for deploying VMs and containers on a consolidated platform. Starting with VCF 4, native Tanzu Kubernetes Grid (TKG) integration provides the first full stack hybrid cloud platform for VMs and containers.
Using Intel’s reference architecture for analytics built with VMware Cloud Foundation with Tanzu, infrastructure teams can build full stack HCI powered by Intel hardware on-premises, at the edge or in the cloud. The hybrid cloud reference architecture is built upon Intel Optane™ persistent memory, second generation Intel Xeon™ scalable processors Intel DL Boost technologies running VMware Cloud Foundation software, delivering an end‐to‐end solution that is ready to deploy, providing the raw performance to power the most advanced AI and machine‐learning workloads that will be a dominant part of our future.
Part 2 of this series will dig deeper into the reference architecture, including a detailed look at the infrastructure building blocks utilized to deploy the Hybrid Cloud Instances, review the benchmarks that are used to model training and inferencing performance, as well as a detailed look at the technologies used to accelerate performance and the specific outcomes achieved. You can download the reference architecture directly from this link.
[1] IDC US46794720 Aug 2020. “WW Spending on AI expected to double in 4 years reaching $110B in 2024”
[2] Diagram courtesy of Intel Corporation – Modernize the Data Center for Hybrid Cloud Oct. 2020
