Cloud computing also often simply referred to as "cloud" represents a group of servers and computers connected to each other over the internet to create a large distributed infrastructure, which can deliver on demand services over the internet on a pay-for-use basis. Most of the cloud computing services could be split into three big categories: Infrastructure-as-a-service (IaaS), Platform-as-a-service (PaaS) and Software-as-a-service (SaaS). The focus of this article will mainly be on IaaS, space dominated by players like Amazon, Microsoft or Google.
Today there is a large spectrum of advantages that drive companies to go for a cloud infrastructure. Some of the most important are:
Lower computation costs - Due to applications running in cloud there is less in-house hardware and support required;
Elastic resources - Scale up or down quickly and easily to meet demand;
Metered service so one only pays for what is used;
Self service - All the IT resources have self-service access.
For many traditional scenarios where you don't have an extreme volume of data, classical cloud computing architecture is covering the demand. But in the era of IoT, Analytics, Blockchain, Machine Learning and AI the amount and variety of data generated as well as the demand for computation of this data will reach unprecedented heights. All this combined with the necessity of taking decisions in real-time will create new challenges for the existing cloud infrastructure.
Therefore, a new technology is emerging, called fog computing, which was first defined as a term by Cisco in order to extend the current cloud computing infrastructure.
Currently there is a fog computing consortium, called OpenFog, which is handling this topic meant to act as the future infrastructure technology.
SONM is part of this consortium, whose aim is to act on the same level as AWS, Azure or GCP, although in core, it has a different infrastructure design, which is decentralized and oriented towards processing massive amounts of data close to the source, where data is produced. SONM has the potential of playing a major role in the future cloud infrastructure and to get some advantage compared to existing players when it comes to future cloud scenarios. Once productive, the SONM platform could serve some interesting scenarios in a single or multi-cloud architecture, like the one below.
On the infrastructure layer of a multi-cloud architecture we could have AWS, Azure, SONM and other cloud infrastructure providers, then an abstraction layer like Cloud Foundry or Kubernetes and then on top of that different PaaS solutions like SAP Cloud Platform which delivers different technologies and products as a service, e.g. Blockchain-as-a-Service, IoT, Machine Learning, Applications, etc.
Let's take for example two use cases, which are growing in demand and importance as we speak.
Many software producers for enterprise are building today an abstraction layer and offering it as a service, called Blockchain-as-a-Service. The purpose of this layer is to decouple the infrastructure layer of different core blockchain platforms such as Hyperledger, Quorum, Ethereum, Lisk, IOTA, etc. from a more generic API offered to customers or businesses in a such a way that they won't need to take care of all the hard topics needed when building a blockchain application. Often the infrastructure component of a blockchain is handled by the BaaS and the user has a cockpit to easily manage this. Of course this a solution which will also mature over time as blockchain itself is a technology still at its early stages. Leaving this at a side, it is in the fundamentals of a blockchain to have a certain degree of decentralization, otherwise it will compete with a standard centralized database, which is currently much faster and cost efficient than a blockchain. Therefore, a BaaS needs to make sure that the nodes run for example on different locations of AWS so that if one location goes completely down, the blockchain is not affected. Generally speaking there are some topics a BaaS needs to handle, apart from the costs of running multiple nodes and the complexity of having a really heterogeneous mix, for example with nodes running on AWS and Azure so that the decentralization goes one step further.
Due to the native decentralization of SONM, this could be one of the first major use cases, running a blockchain platform on top of their IaaS. Besides decentralization the costs will also be much smaller compared to its competitors, as SONM opens the doors for everyone, who would want to share computational power and other ressources such as cloud storage through partners like Storj, thus reducing monopol prices.
IoT, Analytics and Machine Learning
With an increasing number of devices connected to the internet, the volume of data generated and needed to be processed is increasing exponentially. Besides that, new business processes and requirements are emerging e.g. processing the data and taking actions in real-time. For such scenarios the existing cloud infrastructure is not by far an optimal solution, as the data generated by these devices has to be sent to different cloud centers, where is processed and based on that analysis, results and actions are taken by other devices. With the increasing volume of data, the latency between all these processes is becoming a huge impediment in taking actions within a relevant time. To make it clearer let's take the following example: in case of autonomous cars it would be a major disaster if the data would have to be sent to a cloud provider, processed there and then the result sent back to a device, which takes action. Therefore, in case of autonomous cars the data is processed locally, close to the source so that actions are taken in real-time, and only certain data is sent to the cloud for historical analysis and longer-term storage. An autonomous car can basically be seen like an on-premise landscape, where all the necessary infrastructure is centralized in one place, where everything happens. While the example is reflecting the requirements, it does not follow the general cloud strategy and wouldn't be an easy scalable and cost effective solution for large IoT device networks. This is where fog computing comes in place and where SONM could play a major role. With SONM cloud infrastructure the data is processed near the place is produced. Therefore acts on IoT data in milliseconds or faster are possible and new business cases are supported in such a way so that they bring more value.
SONM and fog computing could provide the infrastructure not only for one autonomous car but for a network of connected cars. Through real-time analytics, learning, prediction and adaption, devices will be able to learn and change the behavior dynamically. A network of connected cars connected in real-time to an analytics engine could provide and train algorithms which could predict the future of traffic situation at the current rate. In this way, other cars could be warned about such situations and redirected over other routes, if possible, before a traffic jam is built.
By creating low-latency network connections between the IoT devices and computational endpoints, and offering a large and flexible computational power for massive volume of data at lower costs than existing cloud providers, SONM and fog computing technology could offer organizations more choices for processing data and enable the change of many processes and operations as we know them today.