K
Keith Sudbury
Guest
Quite a bit of your options are going to depend on how your existing Type I areas are set up (tables/indexes per area) , your available hardware/OS and if you really need to do everything at once or can break into multiple maintenance windows (doing selected tables during each window). The short version is multi threading the process as much as possible and avoiding any dead times. Like waiting for the dump to completely finish before starting the load. In most cases I have had better luck with the following overall approach: 1) Running multiple proutil dump processes at the same (some with threads and some without) with the server running with a large -B. Large -B and server running can be skipped if you don't have a lot of mixed table areas. 2) Loading the data into the new database with all of new the extents allocated, -i and reasonable startup settings. Don't build indexes as you load. 3) Running a single multithreaded idxbuild per area once the load is completed for that area. The big variations are going to be tables/indexes that are meaningful percentages of the database and if you are dumping and loading on the same hardware. It isn't easy to come up with a generic solution that works for every database/server but these have served me pretty well over the years.
Continue reading...
Continue reading...