The explosion that Artificial Intelligence is experiencing in different areas, from financial resource planning to cybersecurity, is leading experts to propose the use of intelligent algorithms in all types of ICT areas, being that of network infrastructure management in companiesone of the most promising.
What has been dubbed “AI-Driven network automation” is a set of technologies that combine the power of artificial intelligence and machine learning with the management and operation of telecommunications networks and their underlying systems. On paper, the implementation of these systems offers a series of very interesting advantages for companies, such as network optimization that, by using AI algorithms to analyze data in real time, can improve bandwidth management, identification and mitigation of bottlenecks or improving service quality.
AI Driven Network Automation allows you to automate routine network management tasks, such as configuring and provisioning network devices; makes it easy to identify unusual traffic patterns that could indicate security threats and can even analyze and diagnose problems in real time. Also and in a delegated way, it is capable of allocating resources more efficiently on the network, which can be especially valuable in cloud and virtualization environments and integrates well with technologies such as SDN (software defined networking) and NVF (virtualized network functions).
With different levels of integration, most manufacturers offer their customers AI solutions that aim to improve the automation of companies’ network environments. This is the case, for example, of Cisco and its Cisco DNA platform, Juniper Networks (Juniper Apstra, Juniper Miss), HPE (HPE InfoSight), Extreme Networks, Arista Networks, etc.
Most of these platforms work in a similar way. Upon implementation, they begin by collecting a wide range of data related to the organization’s network infrastructure, which typically includes traffic data, device configurations, event logs, performance measurements, security data, etc. Based on the analysis of that data (often in real time), machine learning algorithms can identify patterns, trends, anomalies…etc. and make automated decisions to optimize network performance and improve network security.
Challenges of AI in network management
But although AI Driven Network Automation has on paper all the potential to streamline the work of networking professionals and easily improve the performance of any corporate network, it is worth remembering that the current development of AI in the field of management Networking is still in its early stages and is not without its problems.
First of all, it must be taken into account that the data quality is essential for the success of this type of solutions, so if the data collected is inaccurate or “noisy” companies run the risk of AI algorithms making incorrect or suboptimal decisions, which can affect performance. It is also worrying that as they are deployed, there is a risk of having excessive dependence on AI, so if it fails or makes incorrect decisions, not only can the network be affected, but it can be difficult for experts to are able to identify the root cause of the error or the origin of an infrastructure malfunction. In this sense, as automation becomes more sophisticated, there may be a loss of human control in decision making, which, as we noted, can become a problem if automated decisions are not transparent or understandable to administrators. grid.
Taking into account both the benefits that this technology can certainly provide and the potential risks associated with it, a good practice is to clearly identify the objectives of automation and after evaluating the best provider or solutions that best fit our needs. , start with relatively simple use cases, such as automating repetitive configurations or detection and resolution of common problems. As success is achieved with initial automation and ongoing monitoring, expansion to other aspects of the network and new areas of operation could be considered.
Going back to the question we initially asked ourselves, is AI Driven Network Automation the future of the network? We have seen that at this time the answer is somewhat more complex than a “Yes” or a “No”. (https://santaclaritalanes.com/) Most experts will say “Yes, but…” or “Yes, but not yet” especially in more complex infrastructures. Despite this, these same experts have little doubt that in a few years the implementation of this type of solutions will be the norm in almost all companies.
