Need guidance for importing a large SVN dump into RTC

Hi,

our SVN repository is quite large:
  - SVN dump file is about 105GB
  - we have about 215000 SVN revisions
  - trunk check out is about 12GB
  - we potentially have about 30 branches and tags to import
  - we have about 25 toplevel directories that we want to import in one component each

We made some experiment importing one top level directory from trunk, it took about 12 hours.
We imported the same top level directory from a branch and it took about 6 hours.

Which guidance could you provide in terms of:
  - how to manage a large history
  - whether incremental migration is feasible
  - how to mitigate migration and continuity of service
  - which gain or loss in performance SVN users can expect with such a large code base

Thanks,

Philippe