Nick Upson skriver:
Hi,before putting this idea into the trackers I wanted to make sure its not totally crazy.When gbak finishes the actual data part of the restore it then creates all the indexes, could it do that in parallel based on the number of processors available. It seems daft that I have to wait for each index to be built, one at a time, when the server has several processors doing nothing
Something that I *would* expect to improve performance of index creation would be to somehow make use of the fact that the data to be indexed passes through memory during data restore. Could indexes be created on the fly when the data "flies by" to be inserted?
By this I do *not* mean to create and activate indexes before data restore and then just insert the data. But perhaps go more "under the hood" in the restore process and build indexes in some intermediate form during data restore, and then use that intermediate form to very quickly create the indexes on disk? I'm thinking perhaps the data could be sorted in some temp space in a way that's well suited to build indexes from.
I would assume that this requires exclusive DB access, but isn't this the case during restore anyway?
Kjell -- ------------------------------ Kjell Rilbe DataDIA AB E-post: kjell.ri...@datadia.se Telefon: 08-761 06 55 Mobil: 0733-44 24 64
<<attachment: kjell_rilbe.vcf>>
------------------------------------------------------------------------------ Virtualization & Cloud Management Using Capacity Planning Cloud computing makes use of virtualization - but cloud computing also focuses on allowing computing to be delivered as a service. http://www.accelacomm.com/jaw/sfnl/114/51521223/
Firebird-Devel mailing list, web interface at https://lists.sourceforge.net/lists/listinfo/firebird-devel
