Can APDB migration be parallelized for tables with tens of billions of rows?

Support
#1

Hi,

We’re currently running an upgrade from a v24 APDB to v29 using dax_apdb_migrate, and we’re facing a severe performance bottleneck due to the scale of our data.

Our table sizes are:

  • DiaSource: 243 million rows
  • DiaObject: 243 million rows
  • DiaForcedSource: 16.5 billion rows

So far:

  • Updating DiaSource (filling new columns) took ～16 hours
  • The DiaForcedSource update has been running for over 6 days and is still in progress

Given that DiaForcedSource is ～70× larger than DiaSource, we estimate the full migration could take weeks or even months at this rate.

Is there any way to accelerate large-table migration, such as: Enabling parallel / multi-threaded data processing?

Any guidance—or plans to optimize dax_apdb_migrate for large-scale deployments—would be extremely valuable.

Thanks very much!