Three data types companies need to prioritize
As more industrial companies seek to become end-to-end digital enterprises spanning production, office and remote working domains, a key fact should become clear: networks are the backbones for such ambitions. The data running over networks act as digital threads tying operations together across those domains. However, not all data in these threads are equal.
Some data, especially in industrial operations, must be prioritized over others. Otherwise, operational performance and asset availability can be compromised. Worse, costly and potentially life-threatening production disruptions can occur. These can lead to grave injuries, litigation, regulatory fines, missed commitments and tarnished brand reputations. The repercussions of downtime extend beyond a facility and can affect many people, depending on the application, with potential catastrophes such as widespread power outages.
This is why information technology (IT) and operations technology (OT) teams must collaborate to ensure their networks are designed to prioritize certain data over others and provide resiliency and availability to transmit data to where it needs to go.
Computer scientists and programmers distinguish data by whether packet payloads carry text, numbers or some sort of multimedia. Depending on context and criticality, each of the three highlighted data types requires specific network prioritization:
1. Information-based data.
Does the data deliver information within user-acceptable latencies of 100 milliseconds (ms) or more? Is the data thread, from source to receiver, available 99.9 percent of the time or better, with no more than eight hours of downtime during a year?
This data type typically travels over enterprise IT networks and consists of emails, file retrieval and sharing, database requests and application responses. In most companies, voice-over-IP (VoIP) technology also uses information-based data to provide person-to-person communications and group conferencing.
In all these use cases, data packet routing can be “best effort” with packets re-sent as needed to meet quality of service (QoS) specifications. QoS, which is defined by latency, jitter and loss, is managed by network designs using reliable settings and topologies.
Typically, if network hiccups occur, users don’t perceive delays in packet delivery. For example, if a file is sent to an office printer and gets held up for a few seconds, most users wouldn’t know it. However, users of voice and conferencing applications will know if VoIP QoS specs are not being met, which is why its data gets prioritized over other information-based data.
Enterprise IT networks are found in office environments, overlaying plant floors, and warehouses. This lets plant personnel as well as their office colleagues or remote workers securely communicate and access information.
In some cases, information-based data can be transmitted over external industrial networks set up for specific applications and operating in non-real-time over cellular, wide-area networks, and even satellite networks. One example is a remote lift station pumping wastewater to a municipal treatment plant. Because its data is not time-critical to operations, it can send data to a centralized control server in periodic batch modes at low data transmission rates that consume little bandwidth.
2. Real-time data.
Does the data deliver operational information, such as counting, condition parameters, and control commands, with minimal latencies, usually less than 20 ms, on a consistent basis without jeopardizing operations? Is the data thread available 99.99 percent of the time, with no more than 48 minutes of downtime during a year? The real-time data distinction is that it supports time-sensitive production operations.
For example, cyclically executing process programs need input data updated in real time to issue control commands to components. Robots and roaming automated guided vehicles (AGVs) must also have real-time data to do their jobs effectively along with all networked plant safety systems.
Real-time data requires a consistent and predictable, or deterministic, network for continued and uninterrupted production. Additionally, to support the latency needs of real-time data, sometimes in milliseconds, or even microseconds, this data must be prioritized over information-based data. OT networks serve as the backbones of complex, mixed-technology landscapes at the field level―including sensors, programmable logic controllers (PLCs), relays, actuators, valves, instrumentation and other devices.
These components must function with precise, deterministic settings, often in harsh operating conditions. These elements also feed and draw operational data into and from a dynamic, vertical infrastructure consisting of a wide range of data concentrators, signal controllers, edge computers and system-wide control systems.
Packet delays can trigger equipment faults, which can lead to unplanned production downtime and added costs. Plants involved in continuous processing, for example, can take hours to come back up to speed and required temperatures. Feedstocks and work-in-progress may have to be scrapped. Cleaning equipment and plumbing may be needed, too. Unplanned shutdowns also can damage sensitive equipment, which may require service, repairs, or replacements, any of which can add time to bringing production back online.
3. Mission-critical data.
Does the data deliver information required in real time to operate equipment and systems without which potential catastrophes could occur? Is the redundancy supporting these digital threads sufficient to provide 24/7/365 availability with virtually no downtime?
Mission-critical data supports key infrastructure such as public communication networks, the energy grid and its constituent utilities, nuclear plants, oil and gas operations, transportation systems and military applications. These applications must operate around the clock, in real- or near-real time and with 99.999 percent uptime or better.
Reliability, durability and availability of this data type are of the highest importance, which is why its packets must be given highest priority of all data traveling over a shared network. In addition, these networks must be designed with immediate failover resiliency along with sufficient redundancy to ensure resiliency. This way, chances of equipment faults can be minimized, if not eliminated. For example, protective relays in high-voltage, electric power substations are one of the most crucial devices in the substation environment.
Not only does a single relay need to be resilient to the environment, but multiple relays must be able to communicate to one another, process data and operate in real time. These real-time protection schemes are mission-critical, due to the requirement for protection of high-dollar assets, fault detection and power grid recovery as well as other critical operations for the consistent and reliable delivery of electricity to consumers.
If companies looking to become digital enterprises need one mission for their OT and IT teams to collaborate on, it would be this: ensure proper data prioritization on their networks and determine whether the data is information-based, real-time or mission-critical. Doing this will make giant strides toward safeguarding production assets while maximizing their availability and performance.
Jonathan Simpson, industrial networking product marketing manager, Siemens Industry Inc. Edited by Chris Vavra, production editor, Control Engineering, CFE Media, firstname.lastname@example.org.
Keywords: Data acquisition, IT, OT
Not all data is equal: some require greater priority than others.
Operations technology (OT) and information technology (IT) teams need to collaborate and determine if the crucial data is information-based, real-time or mission-critical.
What data does your company prioritize most in its operations?