In AS 2000 I have 10 cubes which I process every day using DTS package. It
runs one after another. so each cube query and build aggregations. Until my
10 cubes are not processed my relational database is slow.

In AS 2005 I would like to design some way so that all 10 cubes can query
first and then start building aggregations. so that my relational database is
only slow while cube is reading data.
It may not be good idea to start all 10 same time. But I like to process
some how so that I can run 1 and 2 cube same time and as soon as 1 and 2
cubes query is done it should start running query for 3 and 4 cube while
building aggregation for 1 or 2.

Is this good design or its not possible Thanks for your advice.

-Ashok

SSAS2005 is more powerful for the processing step.
ou can simply process the measuregroups and specifying the level of
parallelism you want.
if you set this value to 2, so only 2 partitions will be processed at the
same time.
SSAS optimize both the loading and aggregation steps and do it in 1 "big"
transaction.

other option... you can process the data of 2 partitions (processdata
option), then process the aggregations (processindex) of these 2 partitions
later.
but, using this method, if your cubes are on the same database you can't do
in parallel both loading & aggregating tasks, the processing task lock the
SSAS database for the first command and the second command starts only when
the first one is over and committed.

maybe you can restrict the number of connections authorized in SSAS. if you
take a look at the data source options you'll found the "maximum number of
connections" option.
Set this to 2 to reduce the impact on your database.
I have not tested this option, but maybe this can helps you too.

"Ashok" <Ashok (AT) discussions (DOT) microsoft.com> wrote