The future of datacenters is at the edge
Mankind has always had an unfortunate tendency to centralize. Tribes, castles, villages, towns, cities, megalopolises... roads, highways, motorways... ... Spontaneously, as humans we have a knack of creating centralized networks to promote exchanges through a central hub. Since the beginning of IT, this way of thinking has also been applied to computer architecture. This approach is now reaching its limit: like cities, datacenters are becoming more and more saturated and are being overwhelmed by demand. Today, with the rise of the digital era and IoT, traditional computing processes are already straining under the load of more than 26 billion objects already connected to the internet. And this situation is unlikely to improve with more than 75 billion connected objects forecast by 2025.
Bringing computing power to the IoT Applications
It is unrealistic to think that the data collected by the 25 billion objects already in use can simply be shipped to a central datacenter or even to a public cloud for centralized processing. Even if the network had the bandwidth required to handle the flow, latency would remain ugly and that’s not even taking into account the impact of centrally processing the data. In a lot of cases, for example when a real time decision is required, processing data in the cloud or a central datacenter is simply impossible.
Edge Computing was born to address these issues. According to IDC’s definition, Edge Computing is a meshed network of micro-datacenters that can process data locally, where it’s been generated. Qualcomm prefers to call it the "cloud edge”. Others have called it “Fog Computing”. The diagnosis however is the same : The time where all data could be processed centrally is rapidly coming to an end. We are in fact switching paradigm and entering a scenario where data is to be processed locally, either directly by the devices or sensors collecting it or by a nearby "micro-infrastructure".
Nutanix is convinced that these edge servers or edge datacenters will be automated and remotely controlled and will also operate completely autonomously. These "micro-infrastructures", or even IoT devices, will provide all the services developers need to host their applications when they need embedded intelligence. And the solution driving these edges will also be capable of managing information flows in an intelligent and secure way.
Hyperconverging the edge
To cut a long story short, companies need the same simplicity at the edge as they do in their datacenters. In fact, they need a complete and consistent solution able to deploy and manage this new decentralized IT, otherwise they will never be able to industrialize IoT deployments. This new solution should simplify the management of millions of objects, and, above all, provide a complete application platform enabling the execution of containers, serverless functions and offering all kinds of integrated services - including AI - to cater to most application needs. This platform should also be able to seamlessly control and secure the flow of data between edges, as well as between edges and the datacenter or public cloud services. Ultimately such a platform should allow applications to be deployed in connected objects, vehicles, cameras, micro-servers as well as in more traditional infrastructures equipped with servers and storage.
This is precisely what our edge platform, Nutanix Xi IoT has been engineered to deliver: Xi IoT is a complete solution that extends the philosophy, consistency and simplicity of hyperconvergence to the edge to unleash the potential of developers and thus enable companies to innovate faster and deploy their IoT applications at scale. Xi IoT is currently being used by multiple large scale organizations to deploy innovative IoT solutions, to improve their customer experience or to power their Industry 4.0 projects.
