Hyperconverged Infrastructure for AI, Machine Learning, and Data Analytics
By: Scott Aukema, Director of Digital Marketing, ViON Corporation
When you hear the terms “artificial intelligence” (AI) and “machine learning” (ML) (and let’s be honest, if you have even a sliver of interest in technology, it’s difficult not to), hyperconverged infrastructure (HCI) may not be the first thing that comes to mind. However, HCI is beginning to play an important role in high-performance use cases like AI and ML with its ability to capture, process, and reduce vast quantities of data at the source of creation in a small form factor. In this blog, the third in a 3-part series on hyperconverged infrastructure, we’ll examine the role HCI is playing in deploying a complete AI solution. If you’d like to read the previous blogs, you can read about the role that HCI plays in enabling a disaster recovery solution and how it is changing the dynamics for edge computing and remote offices.
Hyperconverged infrastructure at the core of a hybrid multi-cloud model bridges the gaps among public cloud, on-prem private cloud, and existing data center infrastructure, enabling organizations to manage end-to-end data workflows to help ensure that data is easily accessible for AI. As organizations develop their AI/ML strategy and architect an IT environment to support it, the resources needed for a successful deployment quickly become evident. This is where many organizations are turning to a multi-cloud environment to support their AI workloads. A recent study by Deloitte found that 49 percent of organizations that have deployed AI are using cloud-based services1, making it easier to develop, test, refine, and operate AI systems. Hyperconverged infrastructure, in concert with a robust Cloud Management Platform (CMP), can accelerate deployment times, making it easier to stand up and take down an AI environment. These AI services in a consumption model provide the agility and resources needed to stand up an AI practice without making a significant investment in infrastructure and tools. HCI is an essential component of the hybrid multi-cloud environment for AI and ML.
In addition to acting as a catalyst between the data center and the cloud, HCI is well positioned to support edge computing – the processing of data outside of the traditional data center, typically at the edge of a network. Data collected at the edge very often is not being used to its full capacity. IT organizations are looking to hyperconverged infrastructure in these instances to capture data where it is created, compress it and transfer it to a cloud or centralized data center at another site. In many instances, the edge can mean hundreds of locations dispersed throughout the country or the world. Consolidating data from these locations allows organizations to create more complete data lakes for analysis to uncover new insights. By combining servers, storage, networking, and a management layer into a single box, HCI eliminates many of the challenges of configuration and networking that come with edge computing. In addition, organizations can coordinate hundreds of edge devices through a CMP, streamlining management, reducing complexity, and reducing costs. Leveraging HCI for edge computing enables data to flow more freely, whether into a centralized data lake or a public or private cloud environment where it can be used to begin the learning and inference process using the available AI models. Once these models are trained in the cloud, they can then be deployed back to the edge to gain further insights.
Hyperconverged infrastructure can streamline edge computing, enable multi-cloud environments, and act as a catalyst to aggregate data for cloud-based AI applications. For agencies that are geographically dispersed and seeking to leverage data from these disparate locations for a more robust analytics practice, HCI should be considered as part of their overall AI strategy. Since we are still in the infant stages of AI and ML, organizations should strive to be flexible and nimble to adapt to changes. Hyperconverged infrastructure provides that agility.
Hyperconverged infrastructure is enabling applications with a versatile platform, which helps organizations accelerate a variety of use cases. Hyperconverged solutions like Nutanix and Fujitsu’s PRIMERGY are helping agencies simplify deployment, reduce cost, improve performance, and easily scale-up and scale-out. Whether it’s AI, edge computing, disaster recovery, enabling a multi-cloud environment, or any other of a multitude of use cases, hyperconverged infrastructure should be considered as part of an IT modernization strategy.
1 – State of AI in the Enterprise, 2nd edition, Deloitte Insights, October 22, 2018