I have a number of OLAP cubes to be processed. There are a series of steps to
process shared dimensions, then the cube-processing tasks. All the
cube-processing tasks have workflow from last shared-dim task to execute "On
Success", but there are no workflows set up among the cube tasks. They are
all siblings waiting on the last shared-dim task. Will these full-process
tasks execute consecutively? Or might separate threads be launched to execute
multiple cube tasks concurrently?

LMcPhee

They should be free to execute when they want. The package will try to fire
them on startup if they are not constrained. If you want to execute them in
order then you need to use Workflow Constraints ideally.

--


Allan Mitchell
www.SQLDTS.com
www.SQLIS.com
www.Konesans.com


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

The only reason I am considering having the cube tasks execute synchronously
is that it might be more efficient/faster. Plus when I experimented with this
with three cube tasks (all use a few common shared dimensions) I
occasionally get processing errors. I am concerned that having multiple
concurrrent cube tasks running which access common tables/dimensions may be
dangerous practice. Should I just make all the cube tasks siblings in the
workflow and trust AS 2000 to handle it? This would make implementation much
cimpler. To set up all the constraints to make it all synchronous would be a
major headache I think. For each cube there would be 3 separate paths to the
next cube's tasks (do nothing, incremental process, full process) and I would
have to set it up so the 3 paths merge. Repeat for every cube...

LMcPhee

"Allan Mitchell" wrote: