HCI use cases are continuing to increase; read up on the role HCI can play in edge computing as well as in advanced computing scenarios such as artificial intelligence and machine learning.
In 2018 the hyperconverged infrastructure market continued its double-digit growth rate, with the largest vendors in this space, especially Dell EMC, VMware (vSAN) and Nutanix, capturing the largest share of revenue. In 2019, we expect to see changes, some as a result of the success this technology is having in the enterprise space and others due to a larger digital transformation that’s occurring in IT in general. In this article, we look at how HCIs are evolving to support edge computing and how they’re aiming to support higher-performance use cases, like artificial intelligence.
Digital transformation is a term for how companies are striving to gain insight through the use of data analytics, insight that will keep them in tune with their markets and ahead of their competitors. This includes more sophisticated analyses, such as with artificial intelligence (AI), but also faster analyses, in some cases in real time. For many applications this means moving the compute process closer to where the data is captured: at the edge.
HCIs are a good choice for distributed, comprehensive, compute infrastructures that can be deployed in remote locations. In video surveillance, an HCI use case pioneered by Pivot3, an HCI can be a collection point and platform to run image recognition or other software to determine its relevance before sending data back to the data center or to the cloud. Manufacturing companies were some of the earliest users of edge computing, using programmable logic controllers (PLCs) to monitor and control equipment in factories. Now, an HCI cluster can collect millions of data points from the entire production process and do more real-time processing on the production floor.
Automotive and equipment testing generates huge amounts of data from thousands of sensors. This used to be collected and shipped back to a data center. Now, HCI nodes can be put on the test bed or vehicle itself and do some of the analysis on the spot. Military organizations, long-time users of ruggedized compute systems, are adopting HCI technology to provide more distributed compute capability to support battlefield decisions.
Most HCI vendors have tried to address the needs of edge computing with smaller nodes and smaller clusters, many with two nodes or even a single node. Now HCI vendors are producing dedicated “edge nodes.” Pivot3 has a ruggedized node that it designed for the U.S. military but it can also provide field data collection and analysis for commercial companies in harsh environments. We’ll see more HCI models dedicated to edge computing use cases in 2019 from other vendors.
HCI got its start in small and mid-sized companies as a way to consolidate their infrastructures and simplify IT. In the enterprise HCI was mostly used for remote office computing, for VDI and as a compute stack for a specific application or project. Over the past few years, HCI vendors have tried to move upmarket and sell HCIs into the mainstream data center, for more Tier 1 applications.
In 2018 we may have seen a turning point in the acceptance of HCI as a platform worthy of an enterprise’s most important workloads. More companies responding to Evaluator Group’s latest study, expected to be published next month, are saying HCI is appropriate for any workloads, with infrastructure consolidation still the No. 1 use case.
As HCI use cases expand and HCI clusters grow to support larger, more diverse workloads, maintaining consistent performance can be a challenge. This has driven the popularity of all-flash configurations and the increase in flash capacities available. Currently, six vendors, out of the 18 represented in the Evaluator Group HCI Comparison Matrix, offer over 75TB of flash capacity, per node, and one has over 180TB.
Besides increased capacity, in 2018 we saw more NVMe adoption, as most HCI vendors added NVMe support to their lineups. One in particular, Microsoft Storage Spaces Direct, also supports persistent memory caching, using Intel Optane (3D XPoint). While other vendors support Intel Optane in an SSD form factor, Microsoft uses Optane in the memory module form factor. This technology offers near-DRAM latency on the memory bus, with the non-volatile characteristics of SSDs.
Similarly, GPUs have become table stakes in the HCI market. All but two of the vendors on the Evaluator Group HCI Comparison Matrix have models that support GPUs. This technology is a foundation of advanced computing applications, such as AI and machine learning. These kinds of projects typically need to run on a separate infrastructure, like VDI has, making HCI a good choice.
HCI continues to mature and drive the growth for IT infrastructure spending. HCI vendors are adding to the capabilities of their products to support the requirements of digital transformation. HCI use cases are also emerging in the distributed and remote applications associated with edge computing. Enterprise IT has started considering HCI for critical workloads, but as cluster sizes increase, networking infrastructure must be taken into account. This is an area we will discuss in our next article.
