There is increasing talk about network ‘fabrics’ in the context of network and data centre technology. But what exactly are they? Why are they important for today’s network traffic? And are all fabrics created equal? Here is a look at why this network underlay is becoming increasingly critical in large campus networks as enterprises and organisations become ever more dependent upon cloud, AI and other advanced technologies.
What is a network fabric?
Providing connectivity for a small number of workspaces or offices requires a simple network; a few switches and connections and one internet connection are sufficient for everyday connectivity. But for a large or complicated campus such as the Eurometropole of Metz (a large network covering the entire French city with multiple endpoints), or for a data centre, traditional networking models are no longer sufficient to support complex wide area networks.
Today businesses and organisations depend heavily on cloud services and bandwidth-intensive applications. Many also depend on the collation of data from IoT devices, and increasingly businesses across sectors are reliant on AI tools. The recent UN Trade and Development (UNCTAD) report predicts that the global AI market will soar from $189 billion in 2023 to $4.8 trillion by 2033; a 25-fold increase in just a decade. Classic three-tier networks (comprising core, distribution and access networks) risk bottlenecks and traffic disruption when such demand for network capacity is placed on them. However, network ‘fabrics’ are capable of quickly and securely connecting and interconnecting a large volume of end points and every part of the network. The term ‘fabric’ is used to describe the network structure and design of a campus environment because, just like a woven cloth with its multiple threads that interweave seamlessly, it connects all parts of the material together.
The benefits of fabric
The advantages of network fabrics include high-speed, low latency connectivity; communications between servers, storage systems and network devices are fast and consistent and unlike traditional networks there is no need for traffic hierarchy. Another key advantage of using a network fabric is that it supports automation, allowing virtual control, configuration, and monitoring. Network fabrics are service oriented and allow the IT department to roll out new services or reconfigure network segments quickly, without the need to manually reconfigure every device. New policies can be made or applied uniformly across the network fabric, so users and devices have the same level of service regardless of how they connect to the network.
The use of network automation reduces human error and speeds up routine network management tasks. Network fabrics also allow for a much greater level of scalability compared to traditional networks; capacity can be controlled in line with specific business needs and demands for more bandwidth through adding on additional switches or links to the network. Network fabrics can also deal better with network failures or incidents than traditional network configurations. Security incidents can be acted upon rapidly by limiting traffic through network segment isolation, thereby preventing large-scale outages.
Enhanced security
Businesses and organisations rely heavily on their networks for daily operations. Along with providing seamless connectivity, networks need to manage user authentication across a diverse range of users including staff, security personnel, contractors, and guests – as well as IoT devices like cameras, door locks, and environmental controls and secure sensitive data to ensure a seamless user experience.
Multi-technology network fabrics offer end-to-end security in a Zero Trust Network architecture. Some solutions also have automated segmentation for both IT and OT environments and AI-powered analytics to enable reliable, end-to-end secure connectivity for organisations across market sectors.
Centrally controlled vs ‘Controller-less’
Many network fabrics, above all the early solutions, tend to use a controller-based architecture. These fabric solutions have a central controller at the heart of the network which manages the network overlay, configures devices, authenticates new end points and ensures network policies are enforced.
The central controller also monitors the network continuously. Other network fabric solutions are controllerless, which means that instead of a central controller monitoring the network and taking actions centrally, the switches themselves manage their own portion of the network. These ‘fabric-capable’ switches share intelligence on the devices they are running and are able to understand and monitor other switches across the network. Enterprises still use management software to have a complete view of the network but in a controllerless environment, it is the switches that share intelligence and govern data paths of each packet rather than a centralised ‘brain’. Controllerless fabrics have several advantages; they are more easily adapted and upgraded when new technologies emerge and because they have less hardware, their overall infrastructure has a reduced carbon footprint.
As the adoption of new technologies such as Cloud, AI and automation becomes ubiquitous, network fabrics will emerge as the networking infrastructure for metropolitan and large campus networks of choice. Indeed, as bandwidth demands on enterprise networks intensify, the organisations that continue to rely on traditional network infrastructure risk disruptions to the network due to bandwidth restrictions and bottlenecks and critically can also be more susceptible to malicious attacks and cybersecurity threats.
Sébastien Claret
Sébastien Claret is Business Development Director at Alcatel-Lucent Enterprise.
