Before choosing a hyper-converged infrastructure appliance, you need to identify the workloads your organization plans to run and then determine whether the appliance offers the features and capabilities necessary to support those workloads.
As part of this process, a number of factors should be evaluated, such as how easy it is to deploy and manage the system, how the system will be secured, its performance and capacity limitations, and how it integrates with third-party tools and services. Once this information is known, you can determine which HCI appliances will support your workloads in the short term and foreseeable future.
Numerous vendors offer HCI appliances, and it can be difficult to differentiate one from the other. Start by evaluating workload-related factors, such as IOPS, latency, network throughput, CPU cycles and overall processing power. You need to understand how quickly a system will respond to application requests, especially as data loads and the number of users increase.
You must also account for capacities. How much data can the appliance store? How much memory is available to VMs?
When evaluating performance and capacity, consider the technologies behind them. For example, some HCI appliances come with all-flash storage and support the NVMe protocol; others take a hybrid approach and adhere to the SATA standard, which is not as performant as all-flash and NVMe. You should also familiarize yourself with the underlying architecture that supports these technologies, such as determining the density and endurance of a flash drive.
This type of information can help you understand the performance capabilities of the appliance and its overall reliability and availability. Ideally, an HCI appliance should guarantee that there will be no data loss and virtually no downtime -- think 99.999% availability. For this reason, you should assess the backup and failover capabilities of the system and what the vendor promises in terms of uptime and availability.
Ease of deployment and management is another top concern. Maintenance and upgrades should be simple and fast, with operations streamlined and automated wherever possible. For example, many HCI appliances support automatic resource discovery and provisioning that can minimize administrative overhead. You should also factor in what it will take to migrate workloads to the new platform.
You want an appliance whose operations function as part of a single, unified effort that streamlines IT processes as much as possible. The HCI appliance should be a fully integrated system that provides centralized access to all resources and operations, and also has the ability to view and monitor the entire hyper-converged environment, including inside VMs. Some hyper-convergence platforms include built-in reporting features or use intelligent analytic services to help optimize performance and proactively address problems to improve operations even further.
Integral to effective management is interoperability -- the ability of an HCI appliance to integrate with third-party tools, systems and services. To help with this process, the system's infrastructure should be based on standards-based technologies and expose the APIs necessary to facilitate interoperability.
This is particularly important when it comes to cloud technologies, which can take the form of private, hybrid or public clouds. The appliance should support, participate in and interact with the various cloud environments to stay relevant in today's data center.
The system should scale as needed to accommodate fluctuating workloads. Some hyper-converged systems require compute and storage resources to scale together, which limits the workloads they can support and still be cost-effective. Next-generation hyper-converged appliances break this barrier down by enabling compute and storage resources to scale independently, providing much greater flexibility.
Flexibility extends to cluster configurations and the ability to build or modify components to meet specific requirements. With appliances -- as opposed to hyper-convergence software -- there are typically fewer options, especially when it comes to using commodity hardware. You will need to determine the degree to which you're locking yourself into specific configurations and whether commodity hardware is even a consideration for subsequent scaling.
For most IT teams, security will be -- or should be -- their top priority. In fact, security is one reason why an HCI appliance should be fully integrated with extensive visibility into all components. The appliance must include mechanisms for safeguarding data at rest and in motion, with protections built into the compute, storage and network layers. You should be able to apply policies and enforce security configurations to both physical and virtual environments. In addition, the platform's monitoring capabilities should extend beyond performance and system maintenance to include security-related tracking capabilities.
Not surprisingly, the more comprehensive your technical support, the more you typically pay, whether directly or indirectly. Subscription fees might seem fairly steep initially, but you could end up spending much more in terms of IT staff time and lost productivity and business if trouble arises.
Before deciding on a hyper-converged infrastructure appliance, assess the support plans offered by the vendor and what each one costs. Then weigh those factors against the level of expertise in your organization and the degree to which you can afford downtime should applications stop working or essential data becomes unavailable.
Establishing hyper-converged TCO is no small matter. You must take into account not only the cost of the appliance -- whether buying it outright or going the consumption-based route -- but IT staff time, training time, migration operations, support costs, system integrations and similar factors. Other points to assess include what it will cost to scale or upgrade components based on projected business trajectories.
The goal is to identify those HCI appliances most likely to lower risk and help achieve goals. You might be looking for a system that supports a single workload or one that can handle many. In either case, first identify the exact nature of those workloads and what it will take to support them. Only then can you decide which system will enhance and simplify your IT operations while improving application delivery, in the short-term and several years down the road.
Assessing HCI appliances based on use cases
One of the first use cases to entice IT to hyper-convergence is virtual desktop infrastructure (VDI). Although resource intensive, VDI tends to have predictable scalability, with each hyper-converged node capable of supporting a fixed number of virtual desktops. As a result, each node can contain identical hardware, making it easier to estimate requirements and scale systems when needed. In addition, because a hyper-converged node contains both compute and storage resources, the overhead associated with distributed storage is eliminated, resulting in better-performing desktops, especially with the advent of flash storage.
When shopping for HCI appliances, you rely on vendors to provide you information on their products. Unfortunately, product info is often driven more by marketing teams. Even product specifications don't always provide the complete picture needed to make a qualified decision. For these reasons, you should rely on other sources to gather information about the product and vendor.
A good place to start is with product reviews, forums, social media, news articles and any other resources that discuss the product or vendor. Pay close attention to what customers are saying. How happy are they with their systems? What kinds of problems have they encountered? How effectively are problems resolved?
Investigate the hardware and software components that make up the product, such as the hypervisor, processor or storage devices. Components are often used for purposes other than hyper-converged appliances, so you can learn a lot by digging into the individual parts. You should also verify whether the appliance or its components are certified for specific software, such as SAP HANA, and what sort of benchmarks might have been performed for specific workloads.
Don't limit your research to the appliance. What have people said about the vendor's other products and the vendor as a whole? Try to get a sense of the company's overall reputation -- including customer satisfaction and loyalty -- and how likely people are to recommend the vendor's products and services. Pay attention to any customer feedback on support services. In addition, try to learn what current and past employees have to say about working for the company.
You need to consider both the short- and long-term when evaluating appliances and vendors. Does the vendor offer a product roadmap? Is the company committed to its hyper-converged infrastructure line of appliances? And don't forget about the vendor's financial well-being. You might get a great deal on an HCI appliance today, but it will do you little good if you can't scale or upgrade the system or get support a year from now.
Because VDI is such a tried-and-true HCI use case, plenty of vendors sell systems that support VDI workloads. For example, Dell EMC offers the VxRail line of HCI appliances that are based on VMware technologies, such as vCenter and vSAN, and are fully integrated with VMware Cloud Foundation SDDC Manager. The VxRail line includes the V Series, a VDI-optimized platform that contains GPU hardware for graphic-intensive desktops and workloads.
Some organizations also turn to hyper-convergence for backup or disaster recovery (DR). The tightly integrated infrastructure and underlying virtualization layer enable instant VM recovery, making it well suited for backup and DR scenarios. The HCI appliances that support these use cases often have a greater ratio of storage than you might find in an appliance that caters to VDI or other workloads.
HCI appliances based on the Nutanix Enterprise Cloud platform can support backup or DR use cases. The platform includes Nutanix Mine, which is integrated into the native hyper-converged data fabric to provide intelligent tiering and advanced data reduction capabilities. Many vendors offer hyper-converged appliances based on the Nutanix Enterprise Cloud platform, including Dell EMC, Fujitsu, Hewlett Packard Enterprise (HPE), IBM, Inspur and Lenovo. One example is Lenovo's ThinkAgile HX Series, which includes an appliance that can support up to 12 TB of HDD storage with only a single processor.
Another hyper-converged use case is the remote office/branch office (ROBO), sometimes referred to as edge computing. For any edge deployment, it's important to ensure rightsizing and to implement systems that can be maintained and scaled easily. A hyper-converged appliance can be a good fit for edge computing use cases because it's easier to deploy and maintain than traditional systems and can often be managed remotely. In addition, edge scenarios usually don't require large storage capacities and can support smaller clusters, depending on the workloads.
Numerous vendors offer HCI appliances that target ROBO environments. For example, Dell EMC sells the VxRail G Series, which starts with two- or three-node clusters. Another example is Cisco HyperFlex, a line of appliances whose engineering is based on the Cisco Unified Computing System. Cisco offers these appliances in all-flash or hybrid storage configurations. The HyperFlex line includes the Edge series that specifically targets ROBO environments. The Edge appliance is an entry-level hyper-converged system integrated with Cisco SD-WAN, a software-defined wide-area networking offering for managing connectivity across a WAN.
Considerations for modern hyper-converged use cases
As HCI appliances have become more flexible, comprehensive and intelligent, they can accommodate more diverse workloads. For example, IT teams now turn to hyper-convergence to implement their private and hybrid cloud infrastructures. Organizations get the flexibility and automation of a cloud service without the complexity that often comes with cloud computing. The right HCI appliance can ease both deployment and maintenance, while providing the virtualized resources necessary to support the cloud computing model.
More vendors than ever recognize the importance of cloud-friendly HCI systems. NetApp HCI incorporates technologies to deliver private and hybrid cloud infrastructures that offer multi-cloud agility and scalability. The platform incorporates NetApp Kubernetes Service, NetApp Cloud Volumes Service and NetApp Cloud Insights to provide a complete cloud ecosystem that can host varying types of applications by enabling users to scale compute and storage resources independently.
Today's hyper-converged infrastructure appliances are also becoming more adept at handling database management systems, ERP and other applications essential to conducting business. To support these workloads, an appliance must deliver high IOPS and network throughput while providing applications with the right balance of compute and storage resources to accommodate specific workload requirements. At the same time, appliances must be extremely reliable and support a high degree of availability.
Numerous vendors offer hyper-converged systems that include technologies to meet these challenges. Dell EMC offers the VxRail P Series for performance-intensive workloads, VMware vSAN includes built-in fault tolerance and advanced availability, Nutanix Acropolis incorporates MapReduce technologies that support advanced storage capabilities, and Cisco HyperFlex has added an all-NVMe node to enable broader support for business-critical workloads.
HPE SimpliVity is another example of a line of HCI appliances that target these workloads. In addition to delivering the necessary performance and capacity, these appliances offer built-in backup and DR, deduplication, compression, network optimization and all-flash storage. The SimpliVity systems are backed by HPE InfoSight, a cloud-based analytics service that incorporates AI technologies and predictive analytics to optimize customer systems and proactively prevent issues from occurring.
In addition, organizations are turning to hyper-converged infrastructure to support big data, AI, machine learning, deep learning, predictive analytics and other advanced workloads. For example, Pivot3 offers an all-flash, NVMe-based appliance that enables users to scale compute and storage resources independently to better support these workloads. The appliance incorporates a patented erasure coding process that combines efficiency, performance and protection to deliver uninterrupted operations.
Some organizations also turn to HCI appliances for their development and testing efforts. Because virtualization lies at the core of a hyper-converged system, it's easy to spin up VMs on demand to support the fluctuating requirements of a development effort. At the same time, today's hyper-converged infrastructure appliances can handle a greater variety of workloads, making it possible to build a wide range of application environments without the complexity of deploying multiple systems for the various workloads.
Do your research
Choosing the right HCI appliance is no small task. You must understand each system's capabilities and whether it can support anticipated workloads. If you plan to move multiple types of workloads to a hyper-converged infrastructure, you might require appliance models from a variety of vendors. You can't assume one model will address all your needs.
Vendors have begun offering HCI products that are optimized for specific use cases, so you need to pay attention to the differences among them. The key is to ensure that any appliance you select conforms to your workload requirements and not the other way around. Only then can you maximize the benefits HCI can offer your organization.
Dig Deeper on Converged infrastructure management
Five ways the hyper-converged infrastructure market is changingBy: Stephen Pritchard
Top hyper-converged systems and composable infrastructure of 2020
Surveying top hyper-converged Kubernetes container platforms
What are 5 hyper-converged backup best practices?