How Long to Migrate My Data?

Gibberish Printed on White Paper White EdgesI am often asked “How long will it take to migrate our data?” My response is almost always, “That depends on the data.” I am not being intentionally vague, I’m just sharing a truth which many do not understand or want to believe.

For sake of argument, I have a client who has a PDM system which has been their CAD file manager for the past 10 years. That’s a lot of data. There are certainly a lot of files, but think about the records in the database (there are frequently multiple records per file, and often records where no file is involved – like for folders or other logical entities). Consider for a moment the variability among the different types of records.  For files of different types, do they share exactly the same attributes?  Or do the parts and assemblies share attributes (like description, mass, and file name) while drawings and documents share different sets of attributes (like document type and paper size)?

Then I explain to my client that the database also stores the links between the parts, assemblies and drawings and that those links and link types are different from one system to another.  These links and like types are often sophisticated enough to influence the behavior between the CAD software and the PDM system, and if handled improperly, will be a nightmare to fix.  At this point, most clients start to understand the magnitude of the migration, and see that there is no simple calculation that will answer the question, “how long will it take to migrate my data?”

Finally, I ask the client if they revision history is important.  Maintaining revision history compounds the complexity of the data migration by roughly one order of magnitude.  Migrating the revision history of a single object with five revisions is complex, but only slightly more involved than migrating five objects with no revisions (consider things like whether you will bring across the order of precedence of revision, for instance).  The real complication comes when migrating revision-specific structures.  Consider a simple example where an assembly (with three revisions) contains two parts (each with three revisions), and there is one drawing per model (and two drawing revisions for each model revision).  How many records are required to represent this situation?  (Hint: it’s more than 25.)

If I’ve scared you with the thought of migrating your data, good.  Data migration isn’t something to take on lightly.  However, there is a silver lining.  You do not have to migrate everything.  There are a number of things you can do to keep data migration costs down.  For instance, a client can place the legacy system into “reference only” mode and only migrate the current revision of records in the system.  Also, many clients choose to clean-up (rationalize) their messy data before starting a migration (non-uniform data contributes significantly to data migration efforts).

Staples Easy ButtonIf my explanation has made sense, then you’ll agree that there is no “Easy Button” for data migration.  Data migration is complex and takes real planning and skill.  If you keep these things in mind, you will be better prepared for the kind of planning and budget that will be necessary should you decide to switch from one PDM system to another.

Tags: , , ,

Read more posts by

This entry was posted on Monday, September 28th, 2009 at 9:45 pm and is filed under Project Management, SolidWorks Enterprise PDM, Systems Management. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.