Industrial IoT is forcing more networks to the edge and away from central data centers. This is what IT should do now.
Network topology has been straightforward for IT for many years.
The plan has always been to consolidate network workloads on a single network that can be uniformly administered. This strategy simplifies work for IT. It also eliminates the potential for network administrative mistakes or omissions that can occur when a plurality of networks are scattered throughout an organization.
In some cases, however, a single, consolidated network simply doesn’t work when different networks require different quality of service levels. An example would be a healthcare company that requires a very high QoS network for telemedicine and a lower QoS standard for its day-to-day enterprise network.
SEE: Tech projects for IT leaders: How to build a home lab, automate your home, install Node-RED and more (free PDF) (TechRepublic)
Now, with the implementation of industrial IoT technology at the edges of enterprises, there again is a call for a distributed network topology. But why?
Edge technology requires proximate processing and storage
If a company is running a fully automated production facility complete with robotics and other electronic equipment that is tethered to the network, the daily volume created by devices, equipment, communications, data transfers, processing and storage is too massive for a centralized network or for the bandwidth needed for data payload transports. It makes sense to implement a smaller, local network in the plant itself.
Public and centralized networks are not designed for the volume and velocity of data that IIoT produces. The ideal IIoT network is private. It should ideally run on 5G broadband for the control, flexibility and coverage it requires.
How IIoT is forcing IT to rethink networks
IIoT is redefining the types of data that enterprises use, and how networks process this data. For example, an IIoT network primarily transmits and processes unstructured data, not fixed record transactional data.
In contrast, the corporate network processes data that is far more predictable, digestible and manageable. The bulk and the traffic of IIoT data virtually makes it a necessity to implement a single, private, dedicated network to each manufacturing facility for use with its IoT. Security is also a concern, because the networks that operate on the edges of the enterprise must often be maintained and administered by non-IT personnel who don’t have training in IT security practices.
It’s not uncommon for someone on a production floor to shout a password to another employee so they can access a network resource — nor is it uncommon admit for someone on the floor to another individual into a network equipment cage that is supposed to be physically secured and accessible by only a few authorized personnel. It’s also not uncommon for an end user to add a new app to the network without telling anyone.
These and other factors are contributing to an IT rethink of network deployment.
IIoT network best practices
Here are five recommended best practices for IIoT networks:
1. Segmented, private networks for IIoT
There are some very good reasons to deploy edge IoT networks as separate, private, “closed loop” entities. The first reason is the enormous payloads of data and processing that these networks must accommodate. The second reason is security. IoT devices and data streams are major entry points for malware, viruses, ransomware and other security threats. It’s good practice to sequester these networks in their own private domains so they can be locked down before a security infection spreads to other networks.
2. Zero-trust security
The non-IT personnel who are most likely to be running edge networks do not have the same training and awareness of security threats as IT. One way IT can keep an eye on edge networks is to use asset management software and zero-trust networks.
Zero-trust networks only allow access to those personnel who are cleared to use the resources they are requesting. Asset management tracks all network activity, and will flag any new resource or device that has been added to or deleted from the network without permission.
3. Training of para-IT skills to edge personnel
Manufacturing supervisors, production managers and others will need basic training in network maintenance and security so they can administer the networks in their facilities. It’s the job of IT to do this.
Minimally, remote network supervisors should understand basic network metrics and abnormal conditions, when to escalate matters to IT, and appropriate physical and logical security practices for the networks they are in charge of.
4. Ruggedized IoT and networks
Edge networks that are in hostile environmental conditions must be ruggedized for these environments. On the network side, it might make more sense to cable networks than to go wireless, since cabling is insulated from outside environmental forces and better able to transport data. IoT devices and equipment, which may have to endure extreme cold or heat, excessive vibration, dust or drops on the floor, should also be tested for their ability to operate under these conditions.
5. Network failover planning
What happens if a network on the edge fails? Do you have a disaster recovery plan for it?
Many companies use a store-and-forward concept, which temporarily stores data on local hard drives and then forwards the data to other downstream servers once service is restored. Nightly and/or periodic midday data backups are also employed. For facilities that operate around the clock, the technology they rely on must be 24/7.
Let the right IIoT software in
If you’re working implementing IIoT within your enterprise, selecting IIoT software is critical. There are hundreds of IIoT platforms and each one is slightly different from the next, so how do you choose? This article — including links to TechRepublic Premium resources — can help.