When I’m thinking about any PLM project, I can clearly see the step when data available in the organization need to be loaded into the system. This step is often underestimated from different standpoints: ability to gather and load information, availability of data definitions, availability of APIs and system performance. I had chance to write before about “legacy data import” as a one of the three major factors impacting mainstream PLM deployment.
I’d make a try to break down legacy data you can face during the implementation.
1. File Legacy
Existing document, drawings, CAD models, Office documents. In most of the cases, these are “un-managed data resources”, that need to be collected, analyzed, imported and stored into the system
2. Relational Databases
In today’s enterprise landscape, lots of data are located into RDBMS system. You can find lots of legacy data here – starting from early dBase tables and going up to various versions database formats and systems Connections to these systems in most of the cases is very straightforward via SQL-compliant driver or software.
3. Computer and Application Legacy
Often, you have systems that were implemented and used or continue to be used by company now. For some reasons, the access of their data storage is problematic. In this case, the only way is to access these applications via an available API or reverse engineer such data sources. Sometime, you can face old, but still functioning computer systems (mainframe is one of the best examples) that continue to operate and keep lots of valuable for organization information.
Import vs. Federation
These are two separate strategies about how to handle legacy data. You can keep the data in the original form and systems. You PLM system will be accessing the legacy data sources to get data, connect and transform it into a new form. The alternative option is to import data in a single shot into a new system. In this case, your legacy data becomes irrelevant, and you move into a new system. It is hard to say what is the best strategy. The situation needs to be estimated and assessed based on the system analyzes. However, I found legacy systems as something that very painful during implementation.
What is my conclusion? Legacy data is important. The amount of data is growing in the exponential manner. To handle legacy data and systems is a very painful task. Each time we come with new systems, the problem of legacy data comes up again. PLM needs to learn to handle foreign lifecycle data or lifecycle data produced by previous versions of PLM systems. It seems to me as a very important functionality that almost missed today. What is your opinion?