Hyperconvergence is the latest stage in the optimisation of infrastructure that has continued to gain pace in the last few years, having been proven to deliver significant benefits.

Allowing for more flexibility and easier management, hyperconvergence is an approach to “cleaning up” data infrastructure. It’s a framework aimed at reducing the complexity of data centres and increasing scalability. It does this by combining computing, networking and storage into a single system. Hyperconverged platforms often include a hypervisor. This is for virtualised computing, virtualised networking and software defined storage. You can cluster together multiple nodes and create pools of shared storage and compute resources, allowing for more convenient consumption.

Hyper-Converged Servers

Broadberry's range of HPC Multi-Node Servers arepurpose-built for breakthrough performance, compute density, high-speed networking and I/O with stability and increased uptime, making it ideal for High Performance Computing (HPC), Storage and Cloud

Quickspecs.

Broadberry TN200B7108-X4L Multi Node Server 12x 3.5

Tyan High Density Quad Node Server for HPC

Form Factor:
Drive Bays:
HDD Size:
Qty Drives:
Drive Interface:
Server Processor:
Memory DIMMS:
GPU Slots:
GPU Support:
Features:: Redundant Power Supply - Optional
Max RAM Capacity:: GB

from: R115,755

Configure

There are a number of benefits behind the implementation of hyperconverged infrastructure.

In the hyperconverged model, the application is the biggest focus. In the past, infrastructure policy has put the focus on the wrong constructs, whereas now most attention is placed on the workload.

Hyperconverged infrastructure lends itself well to a higher degrees of data reduction through deduplication and compression. This leads to less intensive requirements for storage capacity and network bandwidth.

Automation has been impossible for a long time due to many legacy IT infrastructures being so complex and varied. Hyperconverged infrastructure operates on the principles of the SDDC (software-defined data centre), where everything from storage and servers to supporting services are virtualised. Automation of routine operations is possible through the implementation of centralised management tools, like scheduling.

You no longer need to worry about the implementation of hardware from different manufacturers, as everything is included within a single unified environment. With Broadberry servers you can make the most out of this capability as we never vendor lock, leaving you free to take advantage of 3rd party hardware.

Hyperconvergence heavily focuses only scaling easily to bite-sized chunks, which is a huge difference from the previous traditional IT model.

Hyperconvergence is all about the simplification and unification of infrastructure features. Rather than having to manage separate backup and replication products, these technologies are often built into hyperconverged systems.

Due to the node-based architecture, scalability of your hyperconverged data centre is made easy. To match your resource demand, you just add or remove nodes as required

The architecture of hyperconverged infrastructure solutions is a scale out design. As individual nodes are not meant to scale up, you scale out to increase the total memory of your cluster.

Resiliency is a key factor, and hyperconverged infrastructure has been designed with disaster recovery as a major component. It has been created to be resilient to most disaster recovery incidents, often occurring due to one of these scenarios.

Site Loss: A datacentre suffers a severe problem, like a fire or flood.

Equipment Loss: Equipment in a centre becomes damaged.

Component Failure: A set of devices or a single server fails due to problems with hardware or power.

Connectivity Failure: Systems become inaccessible to the outside because of networking issues.

Hyper-Converged Solutions

As more and more companies are beginning to utilise hyperconvergence, there is an increase in the amount of businesses disregarding other infrastructures in favor of this latest infrastructure solution.

Converged solutions are essentially a pre-combined package of software and hardware in a single system for easy management. However, with a converged infrastructure the storage, compute and networking components are still discrete and can be separated. Separation of components in this manner isn’t possible with hyperconverged infrastructure.

Hyperconverged infrastructure has similar goals but goes a step further by focusing less on networking and more on data control and management. It adds further value by throwing software-defined storage into the mix, in addition to adding higher levels of automation and deeper levels of abstraction.

The high level of flexibility of hyper-converged infrastructure allows for more scalable and cost efficient than converged infrastructure. You are able to add blocks of storage and compute as you need. With a hyperconverged infrastructure, if you are in a situation where you need more storage, you can simply snap that onto your stack. Scaling up is as fast as getting another box.

Traditional infrastructure is built out of a number of individual interlinked servers, including storage, networking and backup appliances. Each unit needs to be configured and managed individually. Whereas with hyperconverged infrastructure you have compute, storage and networking combined within a single unit that is centrally managed and bought from one vendor.

Hyperconverged infrastructure is often the cheapest, easiest to manage and least difficult to maintain approaches to the majority of IT needs, making it the best choice ahead of other options like converged infrastructure and traditional IT.

There are two approaches that can be taken when configuring a system, those being whether it will be a single-node system or a multi-node system.

A single-node system only needs a single server or a virtual machine. Whereas on a multi-node system you have multiple servers of VMs.

The advantages multi-node systems feature over single-node systems include better resiliency. If a failure is experienced on a single node, the whole system will be negatively affected. Whereas if a failure happens with a node in a multi-node system, the system can still run on the remaining nodes uninterrupted.

Multi-node systems generally run more efficiently than single-node systems as work is performed across all nodes in parallel.