The TIG Stack in IIoT/OT
Jessica Wachtel /
Apr 19, 2023
Many industrial operators find themselves amid yet another industrial revolution. Deeper insight through artificial intelligence (AI) and machine learning (ML) integrations characterize this fourth wave (or Industry 4.0). Data is no longer just a record occupying server space. It’s alive and providing value. Real-time insights work in tandem with historical records, painting a complete picture of the lifespan of a piece of machinery and/or its components. Unless, of course, a data historian traps that data, preventing it from freely flowing between system integrations, cutting it off from the latest enterprise technology, and boxing it in with software so specific that top talent has no interest in developing expertise.
Data historians are software system remnants of the previous industrial shift when mechanical and analog machinery transitioned to digital, and robots entered the factory floor. Once considered the bleeding edge of technology, data historians now siphon data from the valuable living resources in the modern factory, turning data into stale records in esoteric systems.
Frustration with a legacy data historian isn’t news and it isn’t a solution. It’s just another problem to solve.
Saving money with predictive maintenance
When it comes to maintaining expensive industrial equipment, the common workflow doesn’t cut it. Typically, this involves scheduling regular maintenance both as frequently and infrequently as possible to keep the machinery running properly without draining the maintenance budget. The other situation that arises is scheduling emergency maintenance because of a breakdown. But what if there was a way to use data to forecast when machines need maintenance and schedule repairs when they have the lowest impact on production?
The predictive maintenance approach leverages historical and real-time data with machine learning and advanced analytic tools that can identify patterns and predict machine behavior. For example, historical data provides the details needed to understand the lifespan of machinery. Combine that with real-time analytics to schedule maintenance only when equipment shows signs of wear and nears the end of its lifespan.
Legacy data historians lack the capabilities that make predictive maintenance possible.
Enter The TIG stack: Telegraf, InfluxDB, Grafana
The TIG stack is an open-source alternative to legacy data historians. Telegraf solves data collection challenges, InfluxDB stores and manages time series data, and Grafana enables visualization of that data. Let’s dive deeper into each part of the stack.
Telegraf: Data collection
Telegraf is lightweight, written in Go, has no external dependencies, and requires a minimal memory footprint. You can install it on the smallest devices and it can handle the scale and volume of industrial data.
Telegraf’s architecture is plugin-based. There are over 300 plugins available for just about every service and protocol. Mix and match the plugins needed for your application and IIoT specifications. Because Telegraf is open-source, if a plugin doesn’t exist, you can always create one for an even further customized experience.
There are four types of Telegraf plugins:
Input: Data starts at the input plugins where Telegraf connects to virtually any data source, including MQTT, Kafka, ModBus, OPC-UA, SNMP, GNMI, and more.
Processor: Processors transform, decorate, or filter the raw metrics. Processors can function as data cleaners.
Aggregator: Aggregators apply a desired operation to data (i.e. mean, min, or max).
Output: These plugins push data from Telegraf to a destination source.
InfluxDB: Data storage and querying at scale
If you don’t clean your data in transit, InfluxDB can clean and aggregate data, and more. The benefits of InfluxDB over a data historian don’t end with storage and querying. InfluxDB offers advanced, real-time analytics, can leverage large datasets for AI/ML integrations, and aids with the deployment of predictive maintenance procedures.
InfluxDB Cloud, powered by IOx, is a purpose-built database for time series data. Built in Rust on top of Apache Arrow, this columnar database separates compute and storage for greater flexibility. It uses Apache Arrow Flight SQL to transfer data between clients, servers, and other systems and tools. Many data scientists, data warehousing, and big data engineers are adopting the Arrow ecosystem, simplifying integrations with InfluxDB.
InfluxDB Cloud supports native SQL queries and includes a SQL query, parser, and execution engine. InfluxDB can handle data with unlimited cardinality and users can slice and dice their data across any dimension.
Grafana: Real-time data visualization and alerting
The last piece of the TIG stack brings us back to the factory floor. Grafana dashboards enable you to query and visualize data to provide value in real time. When combined with historical data and predictive maintenance processes, Grafana can show the current state of things versus their expected state. Real-time alerts help bring focus to where it’s needed. Best of all, Grafana seamlessly integrates with InfluxDB.
There are a few ways of setting up dashboards to suit the needs of a specific use case: ready-made dashboards, custom UI Grafana dashboards, and custom-coded dashboards. There’s also a Flight SQL Plugin that enables users to build reports and dashboards common in BI tools. Grafana can display dashboards on a multi-page or single pane of glass setup. You can configure custom dashboards for all stakeholders, whether that’s machine info for shop floor operators at the edge, a single pane of glass view for business analysts at headquarters, or anyone else.
Getting started with the TIG Stack
The TIG stack bends to your business needs rather than your business conforming to the limitations of a legacy data historian. Custom plugins, advanced real-time analytics, and customized sharable dashboards are just a few features that enable predictive maintenance and the attainment of business goals.
To get started, create a free InfluxDB Cloud account today.