How often do you hear about networking when it comes to analytics projects? Probably not enough, especially when it comes to manufacturing and IoT. This is what to consider.
Interconnected machines and processes are the hallmarks of future manufacturing and will feed into big data and analytics technologies for the next generation of "smart" factories. In order to move real-time data from machines through a data pipeline and into analytics toolsets and dashboards, networks will be needed to propel the data.
Networks rank high on IT decision makers' priority lists, yet the internal work of corporate IT remains focused on the management of network traffic that supports everyday person-to-person communications. The practice on manufacturing floors was to leave choices about networking topology and machine-to-machine (M2M) interconnections to vendors, but as this dialogue moves into internal ERP and other higher-level office systems that support analytics and dashboards, corporate IT will be involved.
There are two flavors of Internet of Things (IoT) communications in manufacturing environments: an IP-based network that is hard-wired and that interconnects machines on the floor with the ability to move information to the internet; and a more localized communications scheme where devices in immediate proximity to each other communicate through wireless technology like Bluetooth or over wired Ethernet.
Within the context of a local area network (LAN), IT must assure there is adequate bandwidth and quality of communications to enable the M2M communications that keep a manufacturing production floor running.
In Germany's grandiose Manufacturing 4.0 initiative, the vision is that eventually machines will be continuously communicating with each other across broad distances and between multiple manufacturing facilities around the world. In this scenario, localized network communications will continue to be important, but there must also be high quality of service and continuous communication over global internet.
Using global internet presents its own set of problems, because a corporate IT department does not have end-to-end control over it. Sensing the commercial possibilities, global telecommunications companies are seizing the opportunity to provide end-to-end telecom over internet for M2M environments. This is an area where effective big data management plays an important role.
For instance, if your goal is to have a machine in Asia communicate with a machine in Europe, you want to facilitate sufficient bandwidth and also a clean stream of data that conveys the essence of what is to be communicated. You don't want excess machine "chatter" or log data to bog things down, so data cleaning in real time has to occur.
On the other hand, there are by-products of machine sensor emissions that should never go into the content of M2M communications, but that are useful because they can predict when specific machines are likely to fail, or when it is time to maintain them. Some companies are using M2M input for preventive maintenance that precludes outages in manufacturing that can delay production.
What IT should be considering
To use a dual approach that enables agile and economized M2M communications while ferreting out vital diagnostic information on machines requires sophisticated big data decision making on the part of the network provider, and most likely in a secure but public cloud setting. Here are four things IT should be thinking about in such situations.
- IT should invite network professionals to big data strategy sessions, because big data and analytics can't be facilitated without a quality network pipeline to convey them.
- Network professionals need to set their visions beyond the confines of the LANs that they directly control. Future requirements to globally connect machines and people mean that global telecom carriers and outside cloud vendors must also be involved, and that IT must be able to manage productive relationships with all of these vendors.
- Network professionals need to pay attention to network quality of service (QoS). Today, when you purchase a new hub or router, it comes preconfigured from the vendor with "best practice" default settings for performance. Unfortunately, when you hook all of these devices into an end-to-end network that transports big data or any other kind of data, the network might not be operating at its highest QoS.
To achieve end-to-end QoS over your networks, you will likely have to adjust all of the original OEM QoS settings on your vendor-supplied equipment. Many internal network personnel lack the skills or training for this time-consuming work, so you might want to bring in a network QoS consultant to do it.
- The final touch is getting M2M communications into an end business analytics or manufacturing software that enables key decision makers to see dashboards and reports on manufacturing performance. Most ERP software contains APIs that enable companies to import this data into an analytics engine, but system integration must still be performed in a collaborative effort between the IT network, applications, and system staffs.
- How IoT big data will transform manufacturing automation
- Surge in real-time big data and IoT analytics is changing corporate thinking
- The internet of things and big data: Unlocking the power (ZDNet)
- Ten examples of IoT and big data working well together (ZDNet)
Note: TechRepublic and ZDNet are CBS Interactive properties.