Hyperconvergence: definition, operation and benefits

The term hyperconvergence (or hyper-convergence), often abbreviated to HCI for Hyper-converged infrastructure, first appeared in 2012. At the time, hyper-convergence was seen as the next generation of virtualisation since it offers better performance than traditional converged infrastructure solutions. Nearly a decade later, this prediction is proving to be true as sales of hyper-converged solutions account for over 54.5% of the total converged systems market. However, the specifics and benefits of this type of hardware architecture are still sometimes misunderstood by industry professionals.

So, in this article you will discover everything you need to know about hyper-convergence, how it works and its benefits!

What is hyper-convergence?

Definition

Hyper-convergence refers to a type of computer hardware architecture that closely links several physical server components, in particular the components of 

• storage;

• processing;

• networking;

• virtualisation.

 With hyperconvergence, these critical functions are performed on an integrated software layer. Virtualisation software extracts and aggregates the underlying resources and dynamically allocates them to applications running on virtual machines. It then allows different elements to be combined (or literally, converged) into a single virtual system. For this reason, hyperconverged systems are often confused with converged infrastructure. However, there are fundamental differences between these two architectures.

What is a converged infrastructure?

A converged infrastructure is also based on a set of components (storage, computing units, networking elements, etc.) combined into a single infrastructure solution. This set is most often controlled by a unified management platform. However, unlike hyper-converged systems, the converged infrastructure is not designed as a cluster (clustered network). 

In converged infrastructure, each computing, storage and networking component can be separated from the others. In contrast, in a hyperconverged environment, the components are necessarily grouped together. This clustering is the basis of how hyperconvergence works.

How does hyper-convergence work?

Hyper-converged infrastructures

In concrete terms, hyper-convergence is based on the use of virtual hyper-converged platforms which integrate 

• storage (often software-defined);

• a hypervisor for virtualised computing;

• virtualised networking. 

 These 3 components run on standardised servers (often off-the-shelf and supplied by third party providers). Secondly, unified management makes it possible to have comprehensive visibility of resources, to organise them into pools (of storage or computing systems), to divide them into performance levels, and then to provision them independently of their actual physical location.

Beyond this basic architecture, the operation of hyper convergence can be easily adapted and modulated according to the use cases and needs of a structure.

The use cases of hyper-convergence

The rise of hyper-converged structures is firstly linked to that of cloud computing.

Data centres commonly use this approach because it allows for radical simplification of architectures by reducing the cost and time of data management and application delivery. In a hyper-converged data centre, there is no need for storage arrays, SANs (storage area networks) and dedicated connectivity; servers can be linked together by switches and then controlled through a single interface. This console is itself integrated with the hypervisor management console.

For the same reasons, hyper-convergence is a popular solution for private clouds in companies in all sectors. Hyper-convergence allows IT managers to achieve a level of agility similar to that of a public cloud infrastructure, while maintaining control of their on-premises hardware. 

Finally, HCI offerings are also particularly used in the field of AI, as they make it easier to support machine and deep learning algorithms. Simplification is the major advantage of hyper-converged infrastructures, but it is not the only one.

The benefits of hyper-converged infrastructures

More performance and resilience

In a hyper-converged infrastructure, storage is defined and managed in software. Storage nodes behave as redundant, reliable and resilient storage pools. So if one node fails, the rest of the architecture is not affected. Load migrations to another node are fast and mostly automated.

At this level, the other major advantage of hyper convergence concerns the protection of personal data. With the massive increase in cybercrime, hyper-converged infrastructures often incorporate disaster recovery and automatic backup systems.

Increased flexibility and scalability

The very architecture of hyper-converged infrastructures is based on flexibility. All server, network and storage elements are brought together, allowing to

• reduce the number of components to be maintained and installed;

• increase visibility into performance;

• facilitate the creation of additional nodes;

• simplify migration of workloads from one location to another;

• speed up the configuration of new automated nodes;

• rapidly scale the information system to match the business activity.

More cost-effective than other hardware architectures

By leveraging standard components, rather than proprietary hardware and equipment, hyperconverged systems deliver equal or better performance than traditional architectures – and at a lower cost. These solutions also compete with public cloud services, as they offer the same flexibility and greater control over the hardware.

Finally, hyper-convergence also reduces maintenance costs. Likewise, this architecture requires fewer human resources. For example, in many cases, increasing the storage capacity of a hyper-converged system is as simple as adding flash memory to a server. Other components such as computing systems can be managed at the virtual level from the management console.

Sources and useful links:

Gomindsight’s report on the 5 benefits of hyperconvergence

Networkworld article on how hyper-convergence works