Today, the complexity of datacenter IT environment is continuously growing as the amount of data increases and applications demand more compute power with more infrastructure components needed to support it. At the same time, IT departments should always be able to provision resources instantly while maintaining flexibility and scalability of the infrastructure to handle unpredictable data growth.
Traditional IT infrastructures are comprised of separate compute, storage, and networking components requiring different administrative groups and systems for their management. The storage team, for example, handles the maintenance of the storage subsystem and the relationship with the storage hardware vendor. The same goes for the servers and the network teams. Such infrastructures feature multiple management interfaces for separate components, higher maintenance costs, and are a real headache in terms of support since different components often come from different vendors. All of this makes infrastructure management a highly time- and effort-consuming task forcing businesses to spend their time and money just on keeping the IT infrastructure working instead of focusing on innovations and services delivery.
The problem of the datacenter IT environment complexity and large hardware footprint is partially solved by deploying a converged infrastructure (CI). The main idea behind this concept is to minimize compatibility issues between servers, storage systems, and network switching. To achieve this, a converged infrastructure delivers a set of separate compute, storage, and networking resources optimized and tested for better interoperability.
While reducing the complexity of the datacenter IT infrastructure to a certain extent, such approach has significant drawbacks. The components in a converged infrastructure are managed separately, requiring dedicated applications to manage various pieces of hardware, sometimes making administration a challenging task. Furthermore, such infrastructures have a large hardware footprint – servers, data storage devices, and networking equipment occupy unnecessary space and convert into limited flexibility and scaling options. Additionally, due to a large amount of hardware utilized, CI implies significant deployment and operational expenses.
Hyperconverged infrastructure (HCI) takes this approach to an entirely new level by combining compute, storage, and networking resources in a single “building block” and adopting Software-Defined Everything approach. Software-Defined Storage (SDS) stack eliminates the need for the proprietary storage hardware, significantly reducing costs since dedicated storage hardware is no longer needed. The storage is controlled at the OS or Hypervisor layer with the help of virtual storage controllers. These virtual controllers run on every node within the cluster ensuring unified storage management, better resiliency, and failover capabilities. Software-Defined Networking (SDN) makes network management agile and flexible by providing a centralized interface allowing administrators to manage traffic and quickly distribute network resources where they are needed. It dramatically improves IT infrastructure efficiency by automating provisioning and configuration of the entire networking stack.
Hyperconverged infrastructure simplifies the work of system administrators by allowing them to manage and monitor the entire IT infrastructure from a “single pane of glass”. The hardware footprint is significantly reduced since all components are tightly integrated and delivered in a single building block.
Such infrastructure provides higher flexibility, allowing not only to Scale Up by adding more drives, memory, or CPUs, but also to scale-out by adding additional cluster nodes. More to say, HCI deployment and operational costs become significantly lower with less hardware to purchase and maintain.
Hyperconverged architecture simplifies management by combining all nodes into a single resource pool. This allows managing aggregated resources across individual nodes from a single pane of glass instead of having a set of separate management interfaces for different components.
HCI delivers higher flexibility and allows businesses to scale as needed by adding new modules or nodes. Unlike converged systems, which require considerable investments to scale out, hyperconverged solutions have a much smaller step size, reducing costs for infrastructure development. Hyperconverged systems are also built around the low-cost commodity x86 hardware significantly lowering upfront and operational costs.
Hyperconvergence allows organizations to deploy different applications in a shared resource pool without worrying about the I/O blender which affects VM performance. Hyperconverged infrastructure’s mix of different storage types enables systems to handle both random and sequential workloads smoothly. Since HCI is VM-centric, it allows the system to see through the I/O blender and to optimize individual VMs based on their I/O profile.
Taking into account all its benefits, HCI has become a perfect solution for remote office/branch office (ROBO) scenario. Hyperconverged systems are designed to be self-contained and remotely managed, making them a perfect choice for hosting virtual machines at branch offices. With centralized management capabilities of HCI, there’s no need for specialized IT staff to perform manual operations such as running backup jobs or creating logical unit numbers (LUNs) or quality-of-service policies. Another benefit here is that hyperconverged infrastructures are fast to deploy and deliver a vast amount of resources that are still tightly managed. Network, storage, and compute resources can be deployed at a new site and start working for the company from the very first day.
Hyperconvergence also is suitable for deploying virtual desktop infrastructures (VDI) as it solves all the challenges around VDI. First, HCI scales out by adding additional nodes, simultaneously increasing both compute and storage resources. Provisioning new virtual desktops becomes easier as new CPU, RAM, and storage resources become automatically available to the common resource pool. Further, hyperconverged systems usually feature several data optimization technologies, such as deduplication and compression. As the result, storage density increases since virtual desktops take less space at the storage layer.
