The Manufacturing Integration Platform (MIP) supports interoperability in the Smart Factory. To make this work, the MIP is based on standardized technologies. Bernd Berres, Chief Product Management at MPDV, talks about these technologies in an interview.
Mr. Berres, REST web services is one of the most important technologies used in combination with the MIP. Why is that so?
Bernd Berres: Web services are probably the most widespread technology when it comes to communication between applications. Almost every current programming language and every type of digital system has web services. This technology is perfectly suited to an integration platform such as our MIP. Communication in REST is stateless. In other words, each request stands on its own and does not consider any previous communication, which greatly simplifies matters. Web services are based on the request-response principle, meaning that a response follows each request. This could be just an OK. For more complex requests, the response might be a larger data package that is further processed by the requesting application. A request can be very simple such as a status query or quite extensive like a request to process an extensive data record or save it in the MIP.
The MIP can also communicate via MQTT in addition to web services. What is the difference and why are there these two options?
As I said, with web services you always need a specific request to get a response. By the way, we also call this interface to the MIP the Service Interface. The second important interface of the MIP is the Factory Collaboration Hub (FCH). Here we rely on the MQTT, a lean protocol from the Internet of Things (IoT) environment. In contrast to the service interface, the FCH uses the publish-subscribe principle. This means that the FCH publishes information on specific topics in real time, i.e., whenever this information is generated.
For example, the machine status of a specific machine is always published when it has changed. Publishing means that the FCH sends the information to all recipients who have subscribed to this topic. Each subscriber only receives information they requested and receives in turn information as soon as it is communicated to the MIP. If a machine is down, all subscribers of the machine are instantly informed. If you were to request this information via the service interface, you would receive the information that the machine is down, but only at the time of the request, which can be too late in some cases.
This polling also generates a large overhead because the server has to constantly answer requests even though nothing has changed. It can also happen that during polling the user misses a brief interim status. As the publish-subscribe principle only works in one direction, the two types of communication can also be combined. For example, the FCH distributes the information that the machine is down. Another application can then request information on the currently produced order via the service interface.
Data is stored in the MIP in the form of Business Objects. What are the benefits?
Object-based methods are always at an advantage when a large number of dependencies need to be mapped. The classic list display, as we know it from Excel, cannot keep up with this. Business Objects, as we have defined them for the MIP, can be linked together in any way to form an information model. Business Objects are also described semantically to allow users to read the meaning of the individual properties. As a result, every MIP-based application knows what data is available for an object and how it is to be understood. Neither the user nor the developer has to worry about how the individual instances of the Business Objects are persisted in a database. A set of content access services ensures that all applications only have access to the Business Objects. This type of implementation not only makes handling easier, but also guarantees greater data consistency and security.
The MIP normally runs on a Windows server and uses a Microsoft SQL database. Why is MPDV currently also working with Linux and PostgreSQL?
The MIP is not always the basis for a large IT system. In some cases, manufacturing companies want to start small and use just a few applications with a manageable data volume. The MIP does not always run in the cloud or in the data center. We have a customer where a small edge PC forms the hardware basis for the MIP. Neither a Windows server nor an MS SQL database runs on this edge-PC. What is needed in this environment is a lean base for our platform. The fact that Linux and PostgreSQL are open-source projects also has advantages. Smaller companies in particular do not always have the budget for a cost-intensive server infrastructure from Microsoft. You can quickly implement a use case with an open-source software.
Thank you for a peek backstage of the MIP.
