If you are considering migrating your data to the Windchill PLM, you likely have some concerns. You want the move to go well, after all. You want to avoid time-consuming snags. You want the migration to be as quick as possible. And you want your data to be useful once you make the move. So, here are some tips for migrating your data to Windchill.
Windchill, of course, is the Product Lifecycle Management (PLM) software published by PTC. It promises to help your organization realize value quickly with standardized, out-of-the-box functionality across a comprehensive portfolio of core Product Data Management (PLM) and advanced PLM applications.
“With easy, secure data access for multi-disciplinary and geographically-distributed teams, quality-focused processes, and a data driven approach to manufacturing, Windchill is elevating how product development gets done,” says the company website.
But what you must remember is that Windchill’s “data driven approach to manufacturing” elevates product development only if that data is accurate, complete, and timely. Windchill is only as effective as the data you put into it. If you put poor-quality data in, you will get compromised results on the other end.
Your goal, before migrating any data into Windchill, is to prevent any bad data from making the journey. To do that, you must plan, organize classification models, and conduct up-front data cleaning in a staging environment. Here’s how.
Bad data is information that is either non-conforming, inappropriate, inaccurate, duplicate, incorrect (misspelled or mis-formatted, for example) or incomplete. Consider a typical record, for instance. The product data you work with consists, at its most basic level, of a part number and a description. Your product data might be legacy data, entered into your system a long time ago, or it might be recent.
Your product data might also be a mix of your company data and data from companies you have acquired or merged with. Mergers and acquisitions mean a merging of products—and that means a merging or consolidating of product data. And this is where bad data creeps in. Data over time tends to degrade in quality. Bad data is:
As you consider migrating your data to Windchill, remember that the most frequent source of bad data is insufficient useful information. Lack of meaningful information about a product means you end up with confusing descriptions and misleading data. This makes it difficult to differentiate one part from another.
You have other factors to consider that complicate your data migration plans. Your product data might be stored across multiple locations, in many file formats, and in multiple databases across your enterprise.
That’s bad data. Now, a definition of good data. Good data is clean data. And by clean, we mean data that is complete, accurate, unique (i.e. non-duplicative of other data), and easy to use.
When your data is clean, your products are grouped and easy to find by their characteristics, leading to higher reuse. Duplicate data is easy to discover. Clean data is easier to analyze, too, which helps you with data consolidation, strategic sourcing and other purposes.
Here are three tips for migrating your data to PTC Windchill.
A data migration model within the Convergence Data Design for Retrieval software staging area. (Figure 1)
Are you migrating several systems to one existing system? Or are you doing a new migration? Either way, you need a staging system (Figure 1) that helps you consolidate classification structures, harvest and enrich data, and then normalize and validate your data.
Classification is the organizational structure of the data. If your organizational structure happens to be too complicated or disparate, your end users will suffer. The operational usefulness of your data will also be too complex or disparate for your end users. This is why you must use a staging environment to develop the classification structure until your data works for your company.
You should look for a few key things when creating your staging area. For one thing, allow for search and analysis in the staging data so that you can create detailed models and perform tests on the retrieval of information of that structure. To create the best data structure, consult your end users, with the end result you desire in mind. Determine that end result by determining the reporting and search needs that are driven by the structure.
Another thing to consider is merging your classifications. You might have a classification that works for you, for example, but you need to make a new branch to incorporate new parts from an acquisition or merger of systems. If you simply add to your existing class structures, that can create a mess. But when you create a staging area, you begin with the classification you want to keep today, and use the staging area to modify and play with additional branches of the tree.
To merge your classifications, first load your classification structures using PLMXML, XML, and/or Excel spreadsheets from other systems into your staging area. Then use a flexible staging area to play with the structure until it meets your needs.
You want to get the most value from your classification project. Follow these seven best practices for a parts classification project and you will stay on track.
The next step after building your classification structure is populating the data to its fullest. Remember that the classification structure and search/reporting needs will identify which attributes are key to your data extraction. These key attributes are the ones you must populate as close to 100% as possible. Then, you populate the other attributes with available supporting data.
Next, think of reporting. Your staging area should have reporting that helps you identify attribute population. This is because proper and complete population of attributes is vital to good search and reporting within Windchill. To help with reporting, use staging search and reporting tools to show how the data will look once you have imported it into the final Windchill location.
Finally, after classifying your parts into categories, enrich them with valuable attribute data. Determine the most important attributes for the parts you are clustering. These are your key attributes, the critical attributes that your users will use when searching for the parts they need. Look for key attributes that have good fill rates and that are normalized before you cluster.
Once your data structure is in place, and after you have harvested the attributes, streamline your data for maximum use. Do this in four steps.
Step 1: Normalize – Ensure you have reviewed the data in the attributes and normalized them to standard nomenclature. Otherwise your data won’t show accurately in reports.
Step 2: Validate – Review the data quality and ensure its export readiness by reviewing things like units of measure and field lengths on the attribute fields. Pay particular attention to those fields that are limited in a PLM like Windchill.
Step 3: De-Duplicate – Inevitably, you will end up duplicate parts. But now that the parts are cleaned and attributes are populated, identifying those duplicates and removing them is a lot easier.
Step 4: Export – Once your data passes validations, you are ready to export to Windchill.
To have a successful data migration to Windchill, do the necessary up-front work of data cleansing, validating, and classifying. Organize and clean your data in a staging area before you start your migration process. Always remember to keep the end in mind when cleaning your data. And create a consolidated classification model and harvest key attributes so that the deployment of your Windchill PLM system goes smoothly.
Companies with chaotic or incomplete data trust Convergence Data to scrub that information into an organized, efficient structure. Want to talk with one of our data experts about your situation? Contact us today.