Free DownloadWhat is hyperconverged infrastructure? Guide to HCI
Hyperconverged infrastructure is a software-centric architecture that brings compute, storage and virtualization resources into a single, highly integrated system. Let's take a closer look at hyperconverged infrastructure technology, including use cases and implementation, tradeoffs, current vendors and product offerings and a look ahead to the future of the technology.
A hyper-converged infrastructure can support cloud deployments, but they are not one in the same; the variances in infrastructure should be noted before potential adoption.
Cloud computing has become a mainstay of enterprise operations. Not only do organizations turn to public cloud services for their application and data requirements, but many are now deploying private clouds in their own data centers or other facilities -- perhaps using an open source platform such as OpenStack. But implementing a cloud infrastructure is no small matter and must be carefully weighed against the benefits of public cloud services.
An organization should also consider whether a converged system such as a hyper-converged infrastructure (HCI) might better serve their purposes. Unfortunately, comparing a cloud infrastructure to HCI isn't always a straightforward process, in part, because vendors use the terms cloud and hyper-convergence somewhat loosely and often interchangeably. But the cloud and HCI aren't the same thing. Although they have similar characteristics, they represent different approaches to infrastructure, and organizations should understand these differences if considering them for their data centers.
HCI vs. cloud: What are they?
An HCI platform is an integrated platform that virtualizes compute, storage and network resources and combines them with a software-defined management system to provide a unified platform for hosting virtualized workloads. The platform is made up of multiple server and storage nodes, along with the necessary networking components, to form a single cluster that IT can easily deploy, maintain and scale.
A cloud computing infrastructure, whether private or public, is made up of physical compute, storage and network resources that are integrated into a single architecture. An abstraction layer pools the physical resources and delivers them as services, which applications and users can easily configure and deploy, either through an API or user interface. However, setting up the infrastructure can be a complex and time-consuming task.
Virtualization lies at the heart of both cloud computing and HCI, but virtualization alone isn't enough to define either one. A cloud environment is directly concerned with the user experience, using advanced automation and orchestration to compose the underlying infrastructure. Hyper-convergence has more to do with simplifying IT, following a rigid node-based architecture that greatly simplifies administration but decreases flexibility. Although some HCI platforms have incorporated cloud-like capabilities, they still remain two different approaches to IT infrastructure.
Traditional HCI platforms tightly couple compute, storage and network resources at the node level.
Performance
Comparing performance between HCI vs. cloud infrastructure can be somewhat tricky. With the cloud, it depends whether it's a public platform or on-premises system, and, if on-premises, how the infrastructure is configured. An HCI cluster consolidates all the hardware components into an integrated infrastructure that keeps applications and data close together, offering high network speeds and data rates, while eliminating the bottlenecks that come with a distributed architecture.
In contrast, the physical hardware that makes up a cloud infrastructure can span multiple geographic locations, introducing a wide range of possible bottlenecks. That said, if the components are in close proximity, as might be the case with a private cloud, many of those bottlenecks can be eliminated. In addition, a cloud infrastructure typically collects detailed system telemetry, which can be used in conjunction with the automation and orchestration capabilities to finetune resources on the fly to deliver better performance.
Scalability
The nodes that make up an HCI cluster serve as building blocks for assembling the infrastructure into an integrated whole. The nodes are preconfigured and pre-integrated and can be added with relatively little effort, removing many of the challenges that come with scaling traditional infrastructure. However, an HCI platform has very specific limits when it comes to the maximum number of nodes per cluster or the increments in which resources can be scaled.
On the other hand, a cloud infrastructure can support thousands of nodes. Its distributed architecture, extensive APIs and orchestration and automation capabilities enable it to easily accommodate growing and shrinking the infrastructure as needed. However, scaling a cloud infrastructure is nowhere near as easy as it is with HCI, when taking into account the issues that come with deploying and integrating disparate systems.
A view of a typical cloud infrastructure, which includes servers, applications, clients and other components.
Reliability
An HCI platform comes with fault-tolerance built in. If a host fails, another host can pick up the workload and start running the virtual machines that were on the failed host. The virtualized storage pool can also accommodate drive failures. However, an HCI platform might not include all redundant components, such as adapters or controllers, resulting single points of failure. The platform also operates in a single location and might require additional services or systems to ensure the necessary reliability.
As with HCI, reliability is also built into a cloud infrastructure. The combination of failover services and virtualization help to maintain workload availability. Collected telemetry can be used to alert administrators to problems or kick off automated remediation. That said, the degree of reliability is specific to the implementation. Public cloud platforms tend to go out of their way to ensure reliability, given that their reputations are at stake. Organizations building their own cloud infrastructures must ensure their systems meet their own reliability requirements, even if it means extending the infrastructure to other geographic regions.
Workloads
Hyper-convergence started out as a way to implement virtual desktop infrastructures, but it soon branched out to include other types of virtualized workloads. In addition, some HCI platforms have added support for containerized applications. In fact, HCI increases its versatility with each new generation, making it possible to deploy more diverse workloads than ever.
Even so, HCI isn't nearly as flexible as a cloud infrastructure, which can support a wide range of virtualized, containerized and, in some cases, bare-metal applications. Because of its flexibility and service-oriented architecture, the infrastructure can rapidly reconfigure resources for specific workloads, using the orchestration and automation capabilities as needed. It's self-service features also make it easy for users to quickly deploy the environments they need to run their specific workloads, without requiring IT intervention.
Storage
An HCI platform is, first and foremost, a storage platform. The storage infrastructure is built directly into the platform and consolidated into a common resource pool, which can include SSDs, HDDs or a combination of both. At the same time, the storage capabilities are limited by the vendor's predefined node structure, which determines drive types and scalability. Although most vendors offer a wide range of storage options, these are nothing compared to the flexibility that comes with a cloud infrastructure.
HCI and the cloud aren't mutually exclusive. An organization might choose to deploy a dedicated private cloud on an HCI platform.
Cloud computing enables an organization to deploy just about any type of storage array in just about any location, if the cloud software can effectively communicate with the devices. As with HCI, the storage resources are abstracted and presented as services. However, the cloud takes it a step further by making it possible for users to use point-and-click operations to select the storage they need for their workloads, choosing from whatever storage options are available to that platform.
Data protection
When implementing IT infrastructure, data protection must be a top priority -- not only in terms of disaster recovery, but also when it comes to security and privacy. For the most part, the data protections built into an HCI platform depend on the vendor or the IT team offering that system. However, some disaster recovery is inherent in HCI through its built-in fault-tolerance; although this might not protect against catastrophic events such as natural disasters or successful cyberattacks.
A cloud infrastructure is like HCI when it comes to data protections. The exact protections depend on the implementation. High-profile public cloud platforms take extraordinary steps to safeguard data; however, customers may have to pay extra for services such as snapshots, backups or other disaster recovery features. IT teams deploying their own cloud infrastructures can choose to include these capabilities within their platforms, basing the level of data protection on their specific requirements.
Management
When it comes to management, HCI is the clear winner, bringing with it a high degree of operational efficiency, while streamlining data center operations. Hyper-convergence came into being as a way to simplify the processes of acquiring, deploying, maintaining and scaling infrastructure. Administrators can control components from a single interface, provision resources quickly and easily, and use the platform's API to integrate third-party tools, without needing to bring in outside expertise.
Implementing a cloud infrastructure isn't so easy. Certainly, it can simplify provisioning and automate routine tasks, and it's well suited to DevOps processes such as continuous integration and delivery, but deploying the environment and keeping it running is another matter altogether. IT must acquire, implement and integrate the hardware, install and configure the software, and tie it all together through a secure and highly performant network.
Costs
There is no simple way to evaluate how costs compare between an HCI platform and cloud infrastructure. Both approaches can help cut expenses and inefficiencies, but a true total cost of ownership for either system requires a thorough analysis that takes into account a wide range of factors, including the initial Capex, ongoing resource utilization, IT personnel requirements, user productivity and any other issues that might affect the costs.
The analysis should also consider whether IT will build its own system, perhaps following a reference architecture and which components the organization might already have on hand. In addition, the organization should consider the differences between using public cloud services or implementing a private cloud, if both are being considered.
HCI vs. cloud computing
Hyper-convergence is all about simplifying IT and accelerating virtual workload deployments. It does little to change how we interact with those workloads or the infrastructure that supports them. The cloud, on the other hand, represents a shift in thinking, affecting how we implement applications, store data and deliver services. The cloud puts users center stage so they can more easily access and use resources.
That said, HCI and the cloud aren't mutually exclusive. An organization might choose to deploy a dedicated private cloud on an HCI platform. Although this limits some of the cloud's inherent flexibility, it can help simplify implementation and management, bringing the best of both worlds to a single platform.
