Table of Contents
IT organizations of all sizes are facing several infrastructure challenges. On one hand, they repeatedly receive pressing requirements coming from the business to keep their company swift and proactive, while taking on new digital transformation initiatives. On the other hand, they struggle to keep their budget under control, provision new resources quickly, and manage the increasing complexity while maintaining efficiency at reasonable levels.
A cloud-only IT strategy is not a viable option for many organizations, now more interested in hybrid scenarios from which they can get the best of both worlds. Unfortunately, by mixing cloud and traditional IT infrastructures there is a tangible risk of creating silos, going exactly towards the wrong direction, and complicating the overall infrastructure even more, introducing inefficiencies. Unlike traditional infrastructures, hyper-converged infrastructures (HCI) are a step forward in simplicity and flexibility. HCI enables organizations to hide IT infrastructure complexity while realizing the benefits of a cloud-like environment. HCI simplifies operations and makes the transition of data and applications between on-premises and cloud easier.
Usually based on commodity X86-based hardware and virtualization software, HCI is a software-defined solution that abstracts CPU, memory, networking, and storage devices, organizing them in resource pools. HCI allows the administrator to combine and provision these resources quickly in the form of virtual machines (VMs) and, more recently, as independent storage resources like network attached storage (NAS) filers or object stores. Management operations are also simplified, allowing an increase in productivity levels of the infrastructure while reducing the number of operators and sysadmins per virtual machines (VMs) under management.
Figure 1: HCI Logical Scheme
HCI, originally chosen by small- and medium-sized enterprises as an alternative to their traditional virtualized infrastructures for its simplicity, is quickly expanding its reach to larger organizations while covering an increasing number of workloads. In fact, for the most advanced solutions, the trade-offs on efficiency imposed by simplicity are becoming less relevant.
Figure 2: Workloads covered by hyper-converged infrastructures from the data storage perspective
HCI is a mature market now and solutions can be divided into three categories:
- Enterprise solutions. With a focus on large organizations’ needs, they are characterized by an extensive feature set, high scalability, core-to-cloud integrations, and tools that go beyond traditional virtualization platform management and up in the application stack.
- Small/Medium Enterprise solutions. Similar to the previous category, but simpler and less expensive. The focus remains the simplification of the IT infrastructure for virtualized environments, with limited core-to-cloud integrations and solution ecosystem.
- Vertical solutions. Designed for specific use cases or vertical markets, they are very competitive in edge-cloud or edge-core deployments, but usually, with a limited solutions ecosystem. Often not very scalable, but efficient from the resource consumption point of view, these solutions integrate open-source hypervisors, such as KVM for example, to offer end-to-end support and better prices.
A hyper-converged solution aimed at serving one of the categories described above should include:
- Scalability: HCI infrastructures are based on a scale-out design. Even though modern hypervisors don’t have the limitations they once had, scalability limits can still be found on the distributed storage layer (or virtual shared storage) and in the tools aimed at managing the cluster and additional services. Scalable HCI systems allow the organization to manage fewer clusters and organize resources in larger pools, minimizing failure domains while improving infrastructure availability and overall system efficiency. A fundamental aspect to take into consideration when evaluating HCI infrastructures is the flexibility of the solution in terms of initial configuration, expansion, and additional data storage services. In fact, the ability to configure nodes of different sizes, HW generation, and purposes increases the system lifespan while enabling the end user to respond quickly to his organization’s business needs with the right amount of resources and budget allocation.
- Feature Set: HCI features include the distributed storage layer and management tools with third-party integrations, common across the board. Then, depending on the main focus of the solution and reflecting the categorization defined above, additional features are available on practically every product in the market. Enterprise-focused solutions have a wider range of functionalities with the goal to provide an end-to-end infrastructure stack for enterprises. They expand their reach to data protection, secondary storage, analytics, cloud services, container support, and more. A large feature set is better for simplification of large data center infrastructure and overall total cost of ownership (TCO) but is usually more expensive. On the other hand, HCI solutions designed for small/medium enterprises (SME), cloud service providers, and edge-core/edge-cloud use cases are usually less expensive and are designed to provide the best combination in terms of efficiency and costs for the use cases where they excel.
- Ease of use and usability: One of the main intents of hyper-convergence is to reduce infrastructure friction and complexity. To do so, HCI products should provide tight integration between all the components involved (hypervisor, virtual machines, storage, network, and hardware), leveraging tools and user interfaces that are easy to use, and aimed at reducing the time necessary to perform sysadmin tasks and hiding the complexity behind the scenes. Commercial hypervisors usually offer all the necessary APIs for this type of integration and their management console can be extended with specific plug-ins and UI elements to manage other components of the stack. A different approach is to make the management console of the HCI solution more central, managing all the components from there, including the hypervisor. There are advantages and disadvantages for both approaches but especially with open source hypervisors or multi-hypervisor support, the latter is the most common. Another aspect of usability comes from the sophistication of the entire HCI solution for handling upgrades, system reconfigurations, node decommissioning or repurposing, and so on.
- Flexibility: Reducing infrastructure complexity shouldn’t come at the expense of flexibility and efficiency. In the beginning, HCI was a solution for a limited number of workloads and applications but now, with the right solution, it is possible to cover almost all enterprise needs. HCI flexibility depends on architecture design and how it can take advantage of different resources at its disposal. The ability to manage nodes of different types and sizes, mix hardware of different generations and vendors, taking advantage of advanced load balancing mechanisms for IO operations, are at the base of it. Even more so, the ability to integrate external storage systems or expose resources via block, file, or object interfaces to the rest of the infrastructure can be of great help to limit the initial investment, improve the return on investment (ROI) on HCI, simplify its adoption in heterogeneous environments, as well as create a smooth migration path from traditional to a hyper-converged infrastructure.
- Solution and partner ecosystem: The HCI stack is only a part of the entire IT infrastructure. It usually takes advantage of standard hypervisors, and integration with third parties can be always done at this level, but it is also true that the deeper the level of integrations, the better for the end-users. In fact, by leveraging plug-ins or APIs, a tighter integration can contribute to a better experience for the end-user. This is why many HCI vendors work with partners to build an ecosystem of solutions aiming at extending and improving the functionality of their HCI stack. Some vendors take advantage of the existing large product portfolio in areas like networking or storage, while others are building this ecosystem based on collaborations with other vendors. Areas of particular interest range from data protection to security and cloud.
- Total Cost of Ownership: Albeit HCI solutions are all designed to reduce complexity and therefore improve overall infrastructure TCO, some are more complete than others or with a better focus on specific use cases, allowing them to address TCO issues more efficiently. In general terms, TCO is very difficult to compute but when it comes to hyper-convergence solutions, the parameters described in the previous points all have an impact on TCO. Even more so, looking at the categories defined at the beginning of this chapter, the TCO of an HCI solution also depends on the architecture and its efficiency in running the workloads for which it has been designed.
vendors that are more focused on medium to large enterprises while others are a better fit for smaller organizations or niche use cases, following the classification presented earlier in this chapter. That said, in many cases, the evaluation of an HCI solution really depends on the specific user’s needs and not on the type of organization.