Is hyperconverged infrastructure for me?
2020-05-11 3 min read
Apple has brought an unprecedented level of hardware conformity to the mobile phone market. Despite being criticised by some, this change has allowed for software standards that would have otherwise been impossible.
Previously, developers working with mobile operating systems would need to gamble on which hardware would be best to invest time in. Since the iPhone, even start-ups can be confident of their app’s compatibility with over a billion Apple devices. Today, this compatibility mindset has found its way into data centres.
Classical data centre architecture relies on a clear division between computing, storage and networking. They include a range of devices, from servers to switches to storage, often from different manufacturers. However, data centres full of hardware require significant space and produce large amounts of heat. Providing appropriate cooling increases the already high maintenance costs of running these facilities.
Converged infrastructures were created to improve the efficiency of how hardware was used. Instead of delineating unique tasks for hardware, each machine in the network was expected to be working on something. This minimised the number of idle components and streamlined the whole process.
However, this created a barrier to entry. Converged systems are specifically made and tested for a client, which can take weeks. While possessing aspects of modular automation and easily replaceable components, this approach proved most welcoming to large businesses.
Hyperconverged infrastructure (HCI) takes previous advancements a step further. In the data centre, not only are computing, storage and networking spread across a network, they are now done by the same system, or even the same machine. Each module communicates with the network and completes tasks internally, speeding up the process. Implementing hyperconvergence is also advantageous to users, who can get hold of the required equipment easily and from one vendor.
Adapting quickly to the market is an essential skill in the manufacturing industry and hyperconverged infrastructure comes with scalability. Adding new modules provides incremental growth conveniently and without weeks of waiting, enabling small and medium enterprises (SMEs) to gradually improve their data infrastructure. However, it does come with an added cost. Hardware companies that provide hyperconverged infrastructure, such as Nutanix, promise a 40 per cent decrease in cost of ownership of data centres.
Think it through
Despite the clear appeal hyperconverged systems might have for SMEs, there are things that must be considered before they can be effectively implemented. The reason why costs are reduced, and the system is more adaptable, is the elimination of redundancy. Moving components closer together in a standardised format is key to the system’s efficiency. Any redundancy left in the network, such as external storage centres, could undo this step forward.
Secondly, the price to pay for the convenient installation and maintenance is a lack of customisation of the hardware. A plant manager should be aware of the quirks and traits unique to the model being chosen.
Finally, and most importantly, HCI reminds us of the importance of understanding obsolescence. An incrementally scalable system must be implemented with the awareness that it will eventually need to be replaced. Those turning to new data centre infrastructures must manage obsolescence effectively. EU Automation recommends that every manufacturing company has an obsolescence management strategy, to avoid being caught out.
Not unlike the iPhone, HCI offers an overall more convenient experience for the user. The real difference is in the existing obstacles the infrastructure manages to solve and the potential advantages for growing manufacturers. Manufacturers are therefore evaluating hyperconvergence to see if it is suitable for their applications.