Software-defined networking (SDN) is necessary to allow for the management of machine-to-machine (M2M) data traffic, and cloud and hybrid networks that will underpin the digital revolution, or Industry 4.0, says Riverbed Technology advanced technology group technical director Paul Griffiths.

Centralised network management based on one global business policy is needed to meet predicted demands. This, in turn, will require network-function virtualisation, software-defined wide-area-network (SD-WAN) capabilities and automated storage, compute and device provisioning.

These elements provide the abilities required to effect traffic path monitoring and selection, and classify applications (apps). Deep packet-inspection of the data for apps will also be required. Classification and inspection capabilities allow for integrated and third-party security and the enforcement of app polices, such as path quality and selection and network quality of service, he explains.

The sheer amount of data, notwithstanding typically small amounts of data from sensors, will be staggering, owing to the number of sensors and data-generating processes, he highlights.

“One of the fundamental requirements of any Industry 4.0 network traffic management system will be to recognise the data flows intended for the application that is interacting with the sensor.”

Controlling traffic based on the origin of the sensor will not scale because there will be too many of them. Controlling traffic based on the destination server will also be difficult to manage because the identity and location of the server using current methods – Internet protocol addresses and application port numbers – also do not scale, he details.

“The best way to design the required features and capabilities into networking architecture, as well as the systemic infrastructure as a whole, is through an application-defined approach,” avers Griffiths.

Traffic to and from sensors for those applications that do not require real-time response can be assigned lower priority in favour of critical traffic.

“Knowing the application data and its path through the network provides proactive troubleshooting and a way of implementing security controls, such as encryption and access,” he emphasises.

While M2M communication will add significantly to the traffic transiting the network, a substantial portion of the traffic will end up in data centres. This is because M2M traffic is an excellent data mining and analytics resource.

However, many current SDN implementations are incomplete, complex and do not provide the required visibility. This has led to a patchwork of deployments within data centres that are incompatible.

Further, data used by processes and apps, also called east–west traffic, is not confined to one data centre and the application servers it hosts, and stretches across multiple data centres interconnected by WAN links.

Therefore, Industry 4.0 networks need to have their architecture designed in such a way that capacity and control requirements are easily changed to meet the needs of big data analysis without adding complexity or compromising security. Orchestration needs to be simple and fast, which can be achieved by recognising the apps that are in use, rather than the servers the data reside on, and then managing the networking accordingly, he states.