PLM and ERP integration. Endless topic. My PLM blogging friends – Chad Jackson and Jim Brown decided to discuss it again at their Tech4PD video blog. Even if it is presented as a debate, I found much more similarity between their statements about PLM/ERP integration rather than differences. They both agree PLM and ERP systems are taking a big part of engineering information environment, both systems are important, PLM and ERP are complimentary in terms of their business functions.
There are differences in the positions, of course. Jim is advocating for single source of products data integrated with ERP, CRM and other systems. PLM is the master of product data. From Jim standpoint, the data from PLM system is integrated and represents as product master data to all other data management systems. Chad agrees, the integration needed, but think low-touch integration is more secured and can be less risky. His (Chad’s) opinion is that total integration proposed by Jim is a too big shot to take. Chad advocates for simple integration and so called "one time information push". From Chad standpoint to keep integration simple is an easier and less risky path to integrate data. If you have few minutes, I’d encourage you to watch the video for few minutes (especially if you want to see how Jim takes a swim with polar bears).
So, a simple conclusion after 7 min of video – data needs to be integration between PLM and ERP. Well, it gave me sort of disappointment. Ask anybody in the industry they will tell you about needs to integrated information between PLM and ERP. I like the way Jim present the need – we have to have a comprehensive solution. At the same time, Chad put it down in a very pragmatic way- complex integration won’t work. We need to have something simple in place. The debates between Jim and Chad made me think about possible steps to resolve the complexity of integration between PLM and ERP.
1. Move from "data ownership" to "data publishing". This is an important shift. Enterprise system integration people spent decades debating "data ownership" topic. It is all about "master" and about how to establish a logic of data ownership. The concept of data publishing takes the data ownership upside down. You shift the focus on how to use data. The access can be provided using multiple ways (APIs and variety of data formats). My personal preference to use linked data and RDF/RDFS model (format) for that purpose.
2. Move from "data mapping" to "data description and connection". In other words, link everything. In the original concept of data integration (Chad called it "data push"), the focus is on how to map data located in one system to another system. The whole purpose is to take data in one system, convert (map) and put in another system. After you accomplished "data publishing" concept, you don’t need to transfer data. You just find right data using model qualifiers. It simplify the business logic and helps to establish more reliable data integration mechanisms.
3. Include links to data elements from other systems. Think about this option similar to links on web pages. If you need to extend data or navigate to related data, you just follow the link, which takes you to the right place.
What is my conclusion? I can see PLM and ERP integrations fail in many companies because it relies to old and outdated integration principles. These old principles relies on data ownership and business of protecting data in a boundary of enterprise applications. It is going to change soon. The new one doesn’t move the data while provide an open and reliable mechanism to build an integration system similar to web. Just my thoughts…