not possible.
and the transactions doesn't provide a so huge overhead. (in fact, the
system process into a temporary folder and when its done, the system delete
the old one and rename the temporary to the final name)
so if you have the ability to process the cube without transaction, the time
required will be near the same!
if you want to give access to your users more quickly, try to change your
options to use the lazy aggregation option.
but this cause the cube to become slow until the aggregations are calculated
(its a background process)
so AS load the fact table, build the initial indexes, save everything and
release the transaction; at this point there is no aggregation calculated,
so the cube is very slow. then AS starts a background process to evaluate
the aggregations. (the time required to do this aggregation is longer
because AS will not use all the CPU and priorize the end user access instead
of the aggregation calculation)
another option is to create partitions.
process only the "current month" partition, commit the transaction, the cube
is accessible but only with the current month data, then process the other
partitions and when these partitions are committed, then the user see the
data.
also, try to use better disks to improve your throughput.
"Piotr" <hokah (AT) wp (DOT) pl> wrote
Quote:
Hi,
I wish to achieve such a result:
Backup OLAP -> Start Processing -> IF Processing Failed Restore Backup
I need something like this becouse processing in transaction takes too
long.
regards
Peter |