fotomek/stock.adobe.com

Five hyper-converged infrastructure use cases to watch

We look at workloads ripe for hyper-converged infrastructure deployment. These include virtual servers and desktops, SMEs/remote offices, Kubernetes, and analytics and machine learning

Hyper-converged infrastructure (HCI) offers the promise of simplified IT infrastructure, with the ability to grow easily. It does this by combining compute and storage in a single platform or node that can be built into clusters in a scale-out fashion.

Buyers can pick ready-to-run hyper-converged nodes directly from suppliers, as a package with hardware and software, or build their own using commodity hardware and HCI software.

Gartner defines hyper-converged infrastructure as “a category of scale-out software-integrated infrastructure that applies a modular approach to compute, network and storage on standard hardware, leveraging distributed, horizontal building blocks under unified management”.

In practice, deploying HCI increasingly means a software-first approach, taking advantage of lower-cost commodity equipment. But some higher-performance use cases still lean towards integrated hardware, which usually comes with higher costs.

With software- and hardware-based options on the market, HCI can be an option for a wide range of use cases, including applications that would not justify new hardware.

Five HCI use cases to watch

1. Virtualisation

Virtual servers and virtual desktops are key use cases for hyper-converged infrastructure. Increasingly, software hyper-converged suppliers are focusing on this space.

HCI offers tighter integration and easier management than a system built around separate components.

VMware is currently the largest of the software hyper-converged players, and bases its system on its vSAN storage software. Nutanix, OpenStack and Red Hat are also active in the HCI for virtualisation market.

Virtualisation is also a use case for hardware-based HCI, especially for performance-sensitive applications and where there is an opportunity to deploy new hardware. But hardware-based hyper-converged infrastructure also has a role in virtual desktop infrastructure (VDI).

VDI can benefit from the close connection between compute, storage and the network, but especially from ease of deployment and management. This is particularly the case in new-build deployments. New dedicated infrastructure is easier to manage than retrofitting HCI onto existing server infrastructure and the desktop computers it replaces.

IT departments should, however, ensure they will be able to port their VDI images to other suppliers’ systems in the future.

2. SMEs and remote offices

Hyper-convergence offers an effective way for organisations to improve their management of remote and branch office IT.

Dedicated, integrated hardware might seem an expensive way to equip remote offices, but suppliers realise the potential market for smaller, remote sites where ease of management is important. Companies such as HPE and Nutanix, but also Cisco, have offerings in the market, and costs are falling.

HCI lends itself to environments without a local IT team – the single-supplier approach removes the need to deal with multiple hardware types, and HCI is built for remote management. Suppliers can also preconfigure systems, so the IT team has fewer builds to support.

An IT support service can use HCI’s deployment and admin tools to run a smaller firm’s infrastructure remotely. This applies equally to small and medium-sized enterprises (SMEs) that run their own datacentres where hyper-convergence allows management of all components from one interface.

SMEs are also less sensitive to some of the performance issues that can affect HCI hardware, trading that for ease of use. And the business, or its IT contractor, can integrate local hyper-converged systems with cloud-based resources, for backup and recovery, as well as archiving.

Gartner notes that edge computing is also a growing application for HCI.

3. Containers, and Kubernetes

Containers are a natural fit for HCI, especially in the datacentre. Because containers do not need a hypervisor, they can work directly with the underlying operating system and hardware.

Moving to HCI simplifies the infrastructure further. The most common container orchestrator, Kubernetes, runs on Linux. Applications each run in their own container and share the host server’s resources.

According to Enrico Signoretti, an analyst at Gigaom, businesses can either deploy dedicated HCI infrastructure for Kubernetes, such as Nutanix or Diamanti, a platform optimised for Kubernetes, or use a hybrid model. VMware supports Kubernetes within its hypervisor.

Signoretti suggests many of the reasons HCI works well for virtualisation apply equally to containers. Containers simplify virtualisation and HCI simplifies the hardware.

Dell EMC also offers a tailored platform for Kubernetes, on its VxRail system, and Microsoft is developing its Azure Kubernetes Service (AKS) on Azure Stack HCI. This provides an on-premise alternative to the Azure cloud-based AKS platform for use cases where performance or data compliance prevent organisations from using the cloud.

4. Analytics, machine learning and AI

For analytics, machine learning and artificial intelligence (AI), the drivers for HCI are quick deployment and the ability to scale by adding nodes.

Closely coupling the component parts of IT improves performance and reliability. And the node approach of hyper-converged software works well for applications such as machine learning and AI, where data storage and compute resources often need to grow together to avoid bottlenecks.

Few machine learning or advanced analytics systems are static. Usually, they are designed with growth in mind, with a constant feed of new data. As HCI is designed to scale, it should be able to handle growing data volumes, as well as newer developments – such as streaming analytics – that are compute and network intensive.

Meanwhile, technologies such as Hadoop are designed for distributed environments, and lend themselves to nodes rather than silos of storage and compute resources.

5. Backup and disaster recovery

In some ways, HCI for backup is the simplest use case of all. All it takes is to drop in a second hyper-converged system for redundancy and data duplication. HCI is flexible enough to support backup and disaster recovery systems on-site, at a secondary or failover site, or in the cloud.

HCI is not, however, a storage-focused backup or archiving system. If the requirement is simply to copy data, then other platforms will be more cost-effective. One of HCI’s advantages is its “system-in-a-box” approach. With compute, storage and networking in one place, and built-in support for virtualisation and containers, hyper-convergence is one of the quicker ways to bring a business back online.

Suppliers recognise this, with HPE’s SimpliVity, Cohesity’s secondary storage offerings on HCI, and Nutanix all providing options. Suppliers such as Cohesity are going further and building in archiving to their HCI platform.

So why HCI?  It is an incredibly flexible architecture, especially now that businesses can deploy it through software – to re-use existing assets – and integrate it with the cloud.

And as suppliers develop their products and improve the ability to run conventional and container workloads alongside conventional virtualisation, it is likely to become even more mainstream.

Read more about hyper-converged infrastructure

Next Steps

Top hyper-converged systems and composable infrastructure of 2020

Read more on Converged infrastructure

CIO
Security
Networking
Data Center
Data Management
Close