Industrial IoT data is like raw dirt, mud, and rocks: you know there is gold in there, but you need tools to extract it productively. If you are looking for a great highbanker and dredge combo for extracting gold from IIoT data, you might want to check out Litmus Edge.
A highbanker and dredge combo is a device that can be used for both dredging and highbanking gold from different sources. Dredging involves using a suction hose to extract gold-bearing material from underwater deposits, while highbanking involves using a pump to force water through a sluice box to separate gold from other materials on land. A dredge highbanker combo can switch between these two functions by changing the attachments and settings
I did have to search for the name of these gold extraction machines as I’m not that much of a nerd. Ok, I admit that I do, from time to time, watch the Gold Rush TV series on Discovery. And besides the cameras and the drama of the show, I think there are a lot of analogies between gold digging and Industrial IoT (thinking about it, there is drama in IIoT as well, but that’s a topic for another post or for conversations with my friends Jeff Winter or Rick Bullotta 😊).
Industrial IoT data is like dirt, mud and rocks
Data collected from devices in industrial environments is of a large variety of shapes, forms, and sizes. Legacy and new protocols, many standards and technologies are being used to connect devices, equipment, and assets, resulting in large denormalized datasets that are hard to make sense of. We all know there is gold in there: the ability to harvest, analyze and extract insights from large amounts of data from many different systems used on factory or shop floors opens a whole new range of production optimizations, business models, customer satisfaction enhancements and more. A real gold rush is happening as Industrials have by now well understood the necessity of connecting their assets, IoT-fying (yep, I’ve decided that was a word and forced MS Editor to accept it as such) machines and equipment.
But as they are deploying and scaling IoT solutions, they rapidly face major hurdles. One of the most prominent ones is the heterogeneous nature of the Industrial landscape. Industrial machines and equipment are built to last decades and rely on different communication protocols such as CAN, Modbus, Profibus… And ERPs, EMS, SCADA, DCS, Historians systems all use different standards to represent and exchange information such as STEP, ISA-95, or OPC-UA. This results in a mix of large denormalized datasets that are hard to make sense of. To extract insights from this data, you need large storage as well as powerful compute capabilities.
When searching for gold in dirt, mud, and rocks, you need tools to dig, crush the rocks, filter out and collect gold. It’s the same when searching for insights in Industrial IoT data, you need tools to collect all the data, normalize it, filter out to keep only the relevant and useful data for immediate analysis, store the rest for later, forward it to the right endpoint, services, and apps.
Another similarity between gold digging and Industrial IoT is the necessity to manage the huge amount of data onsite. When digging for gold, you can’t afford to transport all the rocks and dirt for them to be delt with somewhere centralized, that’d be logistically and financially just impossible. It’s the same for Industrial IoT data: there is so much produced on premises that you will not want to send it all raw to the Cloud, you need to deal with it locally, on premises and only send to the Cloud the gems or portions of data that need to be refined.
Introducing Litmus Edge
Litmus Edge presents itself as an Edge Data Platform that allows to instantly connect to hundreds of OT assets and harness the OT data to power insights at the edge.
Litmus Edge is mainly composed of 2 main parts:
Litmus Edge is the platform running at the edge that allows connecting your industrial assets, collecting and normalizing devices’ data, implementing analytics on that data at the edge and sending data and insights to the Cloud.
The Litmus Edge Manager is the Cloud tool that allows you to remotely manage your multiple litmus edge instances at scale.
The diagram below shows a typical deployment topology for Litmus Edge and shows how Litmus Edge can be deployed and connected to different assets in different ways. You can find details in the docs, but in a nutshell, you can integrate Litmus Edge into most Industrial topologies in a secure way, while the Litmus Edge Manager sits in the Cloud from where you can configure and monitor your Litmus Edge instances, without exposing your assets or OT/factory network to the interwebs.
Deployment at the edge
Litmus Edge can be deployed in different ways depending on how and where you want to use it. It goes from installing the software directly on an industrial PC all the way up to deploying it as a docker container or a pod on a Kubernetes cluster. This won’t force you to buy new equipment if you already have hardware on premises that can be used to run Litmus Edge.
In the introduction module of the Litmus user certification, instructions show how to install a trial version on a local Hyper-V or VirtualBox VM but can also use a hosted sandbox that you can have access to for a full week which should give you all the time in the world to try its features for yourselves. I tried both the hosted sandbox and the virtual machine options which offer the same experience once deployed. Note that the setup is mostly turn-key to get to a functional Litmus Edge instance. I didn’t have to go through the network configuration to adapt to its location in a secured network infrastructure, considering I don’t have such infrastructure at hand, but docs are detailed, and I would assume whoever is in charge of such infrastructure would be perfectly fine going through this network configuration.
Data collection and normalization
Out of the box, Litmus Edge supports connecting many types of industrial assets (PLC, DCS, SCADA, Historians and more), whether they are connected over OPC-UA, BacNet, Modbus, CAN, MQTT, you name it. In the Litmus Edge UI (which is exposed by the Litmus Edge runtime itself), you can select from a wide collection of preconfigured devices from all popular device industrial manufacturers.
In the tutorial for the certification, setup is done in the UI, but it can also be done using the terminal or the REST APIs which enables deploying an actual system with many devices programmatically. Deployment at scale can also be done using the Litmus Edge Manger tool describe later in this post.
For each type of device, you can enter its specific settings. For example, for CAN devices, you can enter the Bit Rate, the Interface name, and the Channel.
You can also use the Device Discovery feature to scan the local network for connected devices and get their information.
Once devices are connected, you can start describing their capabilities to help make sense of the data you will receive from them. In Industrial IoT, the notion of Tags is commonly used for that. You can do this manually, uploading a template, browsing devices (for those who advertise their tags), or doing a server search to match tags with a connected device. You can also enrich device’s data with metadata to add context.
You can also go beyond the simple tagging of devices data, using the Digital Twins feature that allows to define models with static and dynamic attributes, rules for transforming data as well as relationships between twins.
In terms of data storage, you can take advantage of an internal or external InfluxDB timeseries database and you can install more Database applications (details later about app integration).
Analytics at the Edge
One of the major hurdles in deploying IoT systems in industrial environments is the requirement to have data not leave the premises and implement analytics as close as possible to the source. But implementing Edge compute is easier said than done. The heterogeneous nature of IIoT assets’ data as well as volumes of data, in addition to the necessity to keep industrial networks airtight and secure all make creating, deploying, and updating workloads at the edge challenging.
Litmus Edge addresses these challenges by making it simple and efficient to create and deploy analytics workloads at the edge through a couple of graphical tools directly in the Litmus Edge UI: a Flow manager and an Analytics functions editor. These tools are built on Node-Red and offer a drag-and-drop experience.
In the Flows manager you can create workflows taking advantage of the many types of nodes for data input and output, simple functions, network, parser, storage, operating system info and dashboard (buttons, gauges, charts…).
To enrich your flows with more advanced Analytics, you can use the Analytics editor.
Analytics edition is based on the concepts of inputs, functions and outputs and the tool offers an impressive collection of analytics functions to use in your flows, including Gaussian filter, moving window average, anomaly detection and more. Furthermore, Litmus added prebuilt KPIs which I am certain will come handy considering there are many KPIs that are common in the industry such as OEE, monitoring asset usage, uptime and downtime, capacity utilization, and more, saving you a lot of time when setting up your system.
As you would expect, you can also leverage Machine Learning at the edge, using existing models, or using connectors to Cloudera, Azure or other ML platforms.
Integrate your favorite application into Litmus Edge and vice-versa
As you can imagine, you can extend Litmus Edge with third-party apps and services. On the Litmus Edge platform itself you can install apps from a marketplace or by adding containers. The platform uses Docker as its app platform which offers a simplified deployment and integration story and allows you to add your own custom apps.
I was able to add the Grafana docker container on my instance of Litmus Edge and setup a simple line chart and a gauge showing real-time data from a device simulator in a matter of minutes following the documentation instructions.
The default marketplace features apps like MongoDB, PostgreSQL, MySQL, Python and ElasticSearch, but you can also create your own private marketplace with your own collection of pre-existing containers or you home brewed ones.
On the other side of the spectrum, you can integrate Litmus Edge itself into your preferred Cloud solution, be it Kafka, AWS IoT, Azure IoT, your own MQTT broker, Google Cloud, … The integration is done through the “Integration tab” on the Litmus Edge UI. Once you have selected the Cloud solution of your choice you can then select the specific topics you want to send up to that Cloud solution. This gives you total control over what data you want to send where.
Litmus Edge Manager
All the operations I discussed previously were made directly on a Litmus Edge instance. To scale these deployments and operations to a real-life infrastructure with several Litmus Edge instances living in different places, you need a centralized tool. That’s what Litmus Edge Manager is about.
The Litmus Edge Manager is not offered as a SaaS solution, which means you have to have your IT department deploy it for you, and it costs $15K/year, meaning, it’s for serious production deployments, not for toying around. So, you can’t try it out for yourself like you can with Litmus Edge. All I could do to learn more about it was to go through the docs and learning content. At first sight, the Litmus Edge Manager seems to have all the features you’d expect from an enterprise tool used to manage a fleet of devices. Not only can you manage many Litmus Edge devices/instances, but you can also implement a multi-tenant solution. You can manage your applications catalog, templatize and automate Litmus Edge deployments, remotely monitor and configure your Litmus Edge instances, and more.
Some more gems for you
Beyond the basics I just discussed, I uncovered some interesting gems that in my opinion attest to Litmus’ relevance in the industry. Delivering on the actual needs and requirements of their customers is front and center in their products, but keeping up with industry standards evolutions and preparing their platform for what’s next is also visible. To learn more about some of the industry standards or trends Litmus is already working on, I recommend reading the following articles:
Get started (and/or certified) today
Litmus recently announced their Litmus Academy. For now, there are only a handful of modules in there, but I am sure we can expect more to be added soon. They also have a certifications section, so you will be able to get certified demonstrating your proficiency in operating or deploying Litmus Edge. When writing this article, the one learning path available is for the Litmus User certification and offers a set of modules for the basics of the solution as well as some more advanced ones. I am a Litmus Edge certified user now :-).
Conclusion
We are past the era of having to develop every part of an Industrial IoT solution, we are now in the era of integration of existing discreet pieces of technology. The same way gold mining is no longer about digging dirt and panning by hand but rather about using the right highbanker and dredge combo, Industrial IoT is no longer about developing all parts of an IIoT solution by hand, but rather it is about integrating solid and productive pieces of technology together. Litmus Edge It is a flexible and powerful tool for extracting insights from your IIoT data. It might feel a bit rough on the edges from time to time, but at the end of the day, it is a flexible and powerful solution for you to focus on the business outcomes without having to waste energy, time and money on developing your own Edge platform.
Comentários