It has been quite some time ago that I read about how NYT was able to make use of Hadoop in conjunction with EC2 and S3 services from Amazon to convert an approx 11 million articles from TIFF into PDF. The same analogy could be drawn when doing a datamigration involving millions of objects and the associated(which can be conted in gigabytes) documents. Usually the way forward would be to use one of the vendor provided migration scripts which would be essentially a half baked wrapper around a crude migration solution, in short a piece of trash with little practical sense.
Leveraging the enormous scale that virtualization can provide would be made feasible by forming a clever data segregation strategy around the inherently inter related nature of PLM data and associated physical files. The idea would be to identify the smallest unit of work(at a part level or even at the BOM level) which can be stateless and then weave the virtualization strategy around that.
I hope this strategy can be feasible in bringing down the execution time by a considerable margin, which can change the way data migration is being done till now.
0 Responses to “Leveraging Virtualization in PLM Data Migration”