Abstraction puts the ‘unified’ in Unified Namespace
The Unified Namespace (UNS) architecture pattern has proven to be an effective means to opening industrial data access up to the entire business, but the road to implementation is not without a few speed bumps.
First, as industrial companies start to establish their hierarchy and build a UNS, they may find it difficult to get their data to follow their own rules. By its nature, a UNS architecture draws from a multitude of different data sources, most of which present data in unique formats. Even superficially similar assets can format the data they generate in completely unique ways, and differences in data generated by wholly different machines, systems, and PLCs are even more stark. To limit problems in creating and operating a UNS, some industrial companies simply publish data from each system and device directly to an MQTT broker in their own topic namespace. This practice is not truly a UNS, and it offers little of the data accessibility and usability promised by this architectural pattern.
Second, the UNS topic space typically follows the hierarchy: Site, Area, Line, Zone, Cell, and Asset. At each level, the information may include data from multiple systems including PLCs, SCADA, MES, CMMS, QMS, ERP, etc. On the consuming side, many users have unique needs that the UNS alone may not be able to meet.
These challenges are what make consistent, easily scalable abstraction a critical part of your UNS.
A UNS can be defined as a consolidated, abstracted structure by which all business applications can consume real-time industrial data in a consistent manner. Today, we want to focus on the ‘abstracted’ portion of that definition and how an abstraction layer prepares any data for all UNS users.
The importance of robust abstraction
One could say that abstraction is responsible for the “unified” part of “Unified Namespace.” The UNS’ abstraction layer ensures that data fits the UNS’ naming convention and hierarchy instead of the underlying sources of the data. That means labeling individual values, combining data from multiple systems, and placing the data in the correct hierarchical position. Within each tier of UNS hierarchy, datasets must be assembled, contextualized, and standardized according to the UNS’ semantics.
Abstraction comes back into play on the data consumer side as well. Some systems simply do not have the ability to consume data presented according to your UNS’ defined semantics, so data must be reformatted according to consumer needs on a case-by-case basis. Subscriber-side abstraction can recontextualize data after the data has been requested, altering the payload from the UNS’ semantic format to whatever the consuming system requires.
The role of Industrial DataOps
A true Industrial DataOps solution can help you create an abstraction layer that contextualizes data to and from your UNS. From the moment the data is generated at the source, the DataOps solution can add the proper context to meet the semantic requirements of the UNS, whatever they may be. Once the data has been requested, the DataOps solution again becomes critical, recontextualizing the requested data on behalf of the subscribed system to ensure the data can be consumed.
Practically speaking, abstraction is best achieved by using data models. An Industrial DataOps solution can be used to generate models that standardize datasets to fit the standards outlined by the UNS—ordering datasets, applying context that adds logical names and units of measure, and, in some cases, even combining data from multiple sources. A more sophisticated DataOps solution will also allow models to be templatized and extended to similar assets for rapid, scalable onboarding.
When data streams are requested from the UNS, the Industrial DataOps solution can again act as an abstraction layer, recontextualizing the data to fit whatever the consuming system needs. While many consuming systems will have no problem ingesting the consistent UNS format, others may have unique requirements—especially those not specifically designed for the industrial market. Using the same modeling capabilities that put data into the UNS format, the DataOps solution can be used to generate unique models for specific systems, ensuring that data is ready for use, no matter the consumer.
A fully-fledged UNS delivers simple, straightforward, and logically organized data access in a browsable hierarchical structure. Publishing data system-by-system into an MQTT broker does not make a UNS and will not solve your data access problems.
To unlock this architectural pattern’s true value, you need to surround it with the right DataOps tools and practices. This approach provides the abstraction capabilities needed to maintain the integrity of the UNS, while creating enough flexibility to meet the diverse needs of the industrial data infrastructure.
About the author
This article was written by John Harrington, the Chief Product Officer at HighByte, focused on product management, customer and partner success, and company strategy. His areas of responsibility include market research, customer use cases, product priorities, go-to-market, and financial planning.