Why web technologies are revolutionizing industrial analytics
Normal HTML was yesterday
But web technologies are much more than static HTML pages with headlines and images nowadays. Their application can be found today in many areas of everyday life. Almost every app on the smartphone uses HTTP(S) requests to get information or send data to a server.
The status quo
In the world of engineering and manufacturing, there are many areas where data can be collected or are already available. Traditional ERP or MES applications are often developed for a platform (such as Windows) and operate one or more databases in the background. In most cases, they are precisely tailored to (corporate) processes and depict the workflow from order to delivery or from production steps to assembly. The collected data is then often only used for documentation. However, for a profound analysis, the programs are most of the time too limited because they are too specific and tied to the presentation capabilities of the platform.
That’s where web technologies can score
Since some years data analysis has not been done with the programming language of a platform such as C ++ on Windows, but more flexible means such as Python or R are used. These tools do not care if they are used to analyze manufacturing or operational data, or if they analyze weather data. They have a wide infrastructure of free and open software packages, that e.g. can also provide the data via an HTTP(S) interface. The so-called RESTful interfaces are an abstraction of these, which allow the transfer, manipulation, deletion or retrieval of data. It does not matter whether they deal with a complex structure as in the case of operating data or simple time series of machine data. Features such as pagination or streaming also allow you to transfer almost any amount of data or split it into smaller parts. This is in conventional data queries e.g. via SQL often much harder and less flexible. However, the various software libraries of the web technologies can use existing systems and read the data firstly via SQL and then pass it on via RESTful service.
Even though the transmission options offered by these technologies are quite flexible, it is often not necessary to transmit the entire database, as this can cost time and memory. In addition, the users’ hardware is not always optimized to perform complex calculations (such as training a prediction model for anomaly detection). These may then better run on a server or a cloud. The above-mentioned programming languages (R and Python) also support the use of graphics cards and specialized software libraries that, when run externally, do not block the end user’s computer.
Once the calculations have been performed, it is sufficient for the user to present him/her meaningful visualizations or to provide intelligently filtered data. Web frameworks in particular can use their strengths here, as they can generate user interfaces quickly and platform-independent thanks to optimized render logics. Software libraries such as Plotly.js or D3 are powerful visualization tools and offer great flexibility for e.g. analyzing manufacturing data. These can be very well integrated into technologies that have replaced classic HTML, such as React.js. Together, much more flexible user interfaces can be created this way than native programming frameworks such as Microsoft’s UWP allow.
Deployment and application
Another advantage over native programming is the delivery of the applications. While classic ERP or MES applications usually require an installation and deposit large files on the user’s computer, web applications can be easily accessed via a browser and are platform-independent, regardless of whether running a Windows laptop, an Android smartphone or an iPhone.
In addition, updates and improvements can be distributed faster and without installation.
Although a cloud application would be feasible, the client-server model of a web application also provides the ability to deploy to a corporate server. This has the advantage that the data and confidential information (for example, the capacity utilization or the operating behavior of the machines) remain in the company.
Through the use of so-called Docker containers, an integration into an existing server network or the subsequent move to a cloud can be completed seamlessly. This minimizes configuration overhead and resource consumption, allowing end users, e.g. an engineer in your own company, the use of features simply by accessing the web page.
If a local use is necessary, apart from the browser, Docker can also help the user to execute the application, if you want to use your own workstation, for example. Another framework that the application can bundle together is Electron. Well-known applications like Slack or WhatsApp Desktop use web technologies to program their user interface and deliver it via electron. You share the code you would see in the browser or smartphone with the one in the desktop app.
Thus, most software distribution scenarios can be realized with web technologies without having to make major adjustments or redevelop the software. Be it on-premise solutions, where the data remains in the enterprise and the manufacturing, cloud solutions, in which no own hardware must be bought, or single applications, which run bundled on a user’s computer.
Since programming simple HTML pages, the world of software development has changed a lot. While some ERP or MES applications require a native implementation, web technologies can leverage their benefits, especially for analysis. Easy integration into the enterprise and easy docking with existing solutions make it even more interesting to use. It is precisely this flexibility that the broad landscape of tools and data sources in the industrial-analytics environment now requires in order to produce productive solutions. In addition, abstracting analytics and rendering data can add significant value to business operations and day-to-day operations in the industry. But users should be able to quickly generate value without great installations and configurations. Lightweight data preparation and visualization using web technologies can play a major role in achieving this. In particular, the rapidly growing developer community and technology maturity in the web environment offer a potential to revolutionize industrial analytics.