Hyperconverged infrastructure is an approach to data center architecture that combines storage, compute, and networking resources in one solution. It’s particularly beneficial for software-defined data centers, which abstract applications and virtual machines from the underlying hardware.
Hyperconverged systems typically run in virtual environments within a data center. Typically, hyperconverged infrastructure uses x86 servers.
Hyperconvergence requires some configuration, and to scale out the infrastructure, administrators must install more nodes. Hyper-converged infrastructure is intended to be highly scalable.
- Difference Between HCI, CI, and Composability
- Top Hyperconverged Infrastructure Providers and Products
- How to Buy an HCI Solution
Difference Between HCI, CI, and Composability
The two other common approaches to data center architecture are converged infrastructure (which came first) and composable infrastructure. Converged infrastructure mainly involves gathering storage, networking, and compute resources in one hardware solution. It is a convenient IT infrastructure for businesses that want pre-configured or pre-installed systems.
Hyperconverged infrastructure has the same goal — combining the three data center technologies into one solution — but it uses virtualization to accomplish this instead.
Composable infrastructure differs from both of these older data center technologies because it disaggregates all three resources. Instead of being bound to hardware or a hypervisor, composable computing, networking, and storage resources connect to the network fabric rather than one server. The composer software moves resources to pools dependent on the need at the moment.
Converged infrastructure helps data centers manage their IT resources in one infrequently changing system. Composability helps data centers manage their fast-changing application and workload needs. Hyperconverged infrastructure is a virtualization solution for businesses that need to scale their computing resources quickly. Hyperconverged platforms work well within software-defined environments, which use virtualization heavily.
Top Hyperconverged Infrastructure Providers and Products
The following companies are major players in hyperconverged and virtual platforms. We have compiled nine of the best products for enterprises searching for HCI solutions.
Nutanix’s Acropolis Operating System (AOS) is an exceptional solution for data center management and high-performance applications. AOS creates a virtual server that can scale when new resources are needed. Nutanix has its own hypervisor, too, known as AHV. AHV supports most Linux and Windows systems.
Clusters of Nutanix nodes are redundant, in case one should fail. All three components — storage, compute, and networking — exist in one cluster. Nutanix offers Kubernetes support. Though users said that it was expensive, they overwhelmingly stressed the solution’s performance for critical applications and the exceptional support from the Nutanix team.
Nutanix provides both good training and support to customers; users cited a learning platform that, though it could be complex, offered videos, including training for those who wanted to customize and configure AOS.
Dell EMC VxRail
For businesses that use VMware technology heavily, Dell EMC VxRail offers multiple integrations with the virtualization software. This includes VSan, another product on this list, which supports HCI environments. VxRail also offers VMware Tanzu, a Kubernetes solution that comes in three different options, depending on companies’ needs for cloud, traditional app development, hybrid cloud, PaaS, or native Kubernetes pods.
VxRail performs extensive testing for the system, validating clusters and testing hardware. Users can quickly deploy virtual machines and environments.
VxRail is also able to run SAP HANA, a relational database management system. Enterprises that have intensive storage needs such as AI, video, and media streaming will benefit from Dell EMC storage products, including VxRail. Dell also recently added disaggregation to VxRail, allowing users to separate storage and compute resources.
NetApp’s hyperconverged solution offers hybrid cloud-style data management for a variety of clouds and workloads. Users can deploy the HCI solution on multiple clouds and on-premises infrastructure.
NetApp’s HCI is intended to create a type of data fabric for a company: it allows businesses access from a variety of locations, and it also integrates with other NetApp technology, such as Cloud Backup. NetApp expands and improves upon the traditional method of HCI by adding flexibility: users can have both their public cloud and their on-premises management.
NetApp HCI clusters are managed by VMware VCenter, and NetApp uses VMware as its virtualization platform. Users can scale their storage and compute resources separately, a helpful money-saving feature that decreases overprovisioning and keeps unnecessary storage costs down.
One of the most popular HCI solutions, Cisco’s HyperFlex hyperconverged platform offers users an option to use only NVMe in its HX220c M5 All NVMe. A mouthful, but it’s the best choice for critical business workloads and applications because of its speed and low latency. The All NVMe solution uses Intel Optane hardware for its persistent memory.
Users find HyperFlex easy to set up and begin using. They have the choice to create multiple data stores for their important data, rather than being restricted to one pool. HyperFlex can be deployed at edge sites, managed by Cisco Intersight, a cloud-based infrastructure and workload platform. Users also have access to Kubernetes through Intersight.
Microsoft Azure Stack HCI
Azure Stack HCI is a cloud-based recovery and monitoring solution that can be run on premises. Azure Stack HCI runs Windows and Linux virtual machines in either a data center or an edge environment. Microsoft calls its hybrid-cloud style service simply “Azure hybrid.” The HCI solution also integrates with other Azure programs, like virtual machines, and all of these Azure programs can be managed alongside the Stack. This includes Azure Backup, which is available to HCI users.
Azure Stack clusters contain between 2 and 16 servers, which run an operating system that is specifically designed for an HCI. Azure Stack also supports Hyper-V. Virtual machines within Kubernetes nodes are designed to fail over: Kubernetes will move any container from a failed environment to a working virtual machine.
StarWind Hyper Converged Appliance (HCA)
With its all-flash, simple hyperconverged system and comparatively low prices, StarWind is the ideal HCI for smaller enterprises that still need exceptional performance and support. A hyperconverged appliance, or HCA, is composed of two servers. Clusters are also composed of two nodes for failover: if one node fails, the other can take over.
StarWind Command Center, a management console that allows users to monitor their HCA performance, has an HTML5 web interface. StarWind uses Intel Optane technology for persistent memory and Dell EMC servers. StarWind reduces TCO and IT management costs, according to the vendor, and also offers free configuration and deployment — this is a big deal for smaller businesses. It’s one of the less expensive HCI solutions, according to users who benefitted both from the features and the lower cost. Users also reported that they received good support from StarWind’s team.
HPE’s all-flash array, acquired in 2017 when HPE purchased storage company Nimble, uses what HPE calls dHCI, or disaggregated hyper-converged infrastructure. dHCI, not quite converged or hyper-converged infrastructure, allows users to scale which aspect — storage, compute, or networking — they want when they want.
Customers also take advantage of intelligent predictive platform HPE InfoSight, which notifies customers when problems arise, applications go down, or the array needs to scale to compensate for growth. HPE InfoSight connects directly to the dHCI stack.
Nimble provides disaster recovery replication snapshots, including for Hyper-V virtual machines. The replication snapshots can be extended to storage arrays in other physical locations.
Pivot3 offers four HCI products in its Data Center Series, two of them hybrid and two of them flash. Pivot3 hyperconverged environments have 8-16 nodes, depending on which of the four solutions you choose. The clusters of nodes in each appliance have PCIe cards.
Among other technologies — SSD and HDD included — Pivot3 uses NVMe flash. Pivot3 prioritizes storage for business critical applications, using QoS policies to distribute workloads where they will perform best. Pivot3 storage and compute can scale independently, allowing users to save money on capacity for either resource.
Acuity uses an HTML5 graphical user interface. It can integrate with VMWare Vsphere.
VMware vSAN is a software-defined storage platform that supports HCI environments. Aggregated data storage devices form a virtual storage pool. vSAN integrates with Vsphere, providing cloud virtualization for servers. The main benefit of using this HCI is its direct integration with other VMware products.
Users can learn to manage and configure their own vSAN cluster. VMware provides a File Service for sharing files within the Network File System. vSAN can be deployed to top public cloud solutions, and it integrates with Tanzu, VMware’s container and VM management solution. vSAN can also be used for disaster recovery.
How to Buy an HCI Solution
When purchasing a hyperconverged solution, consider the following points.
The best HCI solutions should provide maximum visibility — which workloads don’t have the resources they need, when nodes or clusters fail, how virtual machines are performing. A solution that includes built-in analytics or even AI will be poised to provide necessary insights, especially in a tech environment where data centers are automatically expected to have intelligent systems and machines.
Some solutions allow independent storage and compute scaling, but not all do. If you’re looking to really save money, choose an HCI solution that allows users to scale their storage and compute needs separately. Additional storage costs money, and paying for more than you use can be quite costly.
If you know you’ll need critical applications to move between servers or environments, watch for solutions that support containers. Multiple HCI solutions integrate with Kubernetes, for example. Containers are an important part of future data centers — they help enterprises prioritize workloads and reduce slowdowns and outages.