Routing logs to individual log file in multi rules_file MaxL

Question

Hi Gurus,
I have been pretty late to this forum after long time. I have a situation here, and trying to find out the best way for operational benefits.
We have an ASO cube (Historical) keeps 24 months snapshot data and refreshed monthly just like last 24 months rolling. The cube size is around 18.5 GB. The input level data size is around 13 GB. For monthly refresh the current process rebuilds the cube from scratch, deletes the 1/24 snapshot as it is going to add last months snapshot. The entire process takes 13 hours of processing time becuase the server doesn't have number of CPUs to support parallel operations.
Since we recently moved to 11.1.2.3, and have ample amounts of CPUs(8) and RAM (16gb), I'd like to take davantage of parallelism, and will go for incremental load. Prior to that since the outline build is EPMA driven I'd only like to rebuild the dimension with all data, essentially restructures the DB, with data after metadata refresh, so that I can keep my history intact, and only proceed for loading the last month's data after clearing out the 1st snapshot.
My MaxL script looks like below:
/* Set up logs */
set timestamp on;
spool on to $(mxlLog).log;
/* Connect to Essbase */
login $key $essUser $key $essPwd on $essServer;
alter application "$essApp" load database "$essDB";
/* Disable User Access to DB */
alter application "$essApp" disable connects;
/* Unlock all objects */
alter database "$essApp"."$essDB" unlock all objects;
/* Clear all data for previous month*/
alter database "$essApp"."$essDB" clear data in region 'CrossJoin({([ACTUAL])},{[&CLEAR_PERIOD]})' physical;
/* Load SQL Data */
import database "$essApp"."$essDB" data connect as $key $edsUser identified by $key $edsPwd using multiple rules_file 'LOADDATA','LOADJNLS','LOADFX','LOAD_J1','LOAD_J2','LOAD_J3','LOADDELQ' to load_buffer_block starting with buffer_id 1 on error write to "$(mxlLog)_LOADDATA.err";
/* Selects and build an aggregation that permits the database to grow by no more than 300% */
execute aggregate process on database "$essApp"."$essDB" stopping when total_size exceeds 4 enable alternate_rollups;
/* build query tracking views */
execute aggregate build on database "$essApp"."$essDB" using view_file 'gw';
/* Enable Query Tracking */
alter database "$essApp"."$essDB" enable query_tracking;
/* Enable User Access to DB */
alter application "$essApp" enable connects;
logout;
exit;
I am able to achive performance but not satisfactory. So I have couple of queries below.
1. Whether bule shaded codes can further be tuned. I have major problem in clearing only 1 month snapshot : where I require to clear one scenario and the designated 1st month.
2. Multiple rules_file statement, how do I write logs of each load rule to separte log files instead one, my previous process is wrting error-log for each load rule in separte log file and consolidates at the end of batch run to a single file for the whole batch execution.
Apprecaite any help in this regrad.
Thanks,
DD

Answer 1

Thanks Celvin. I'd rather route MaxL logs in one log file and consolidate into the batch logs instead of using
multiple log files.
Regrading Partial Clear:
My worry is, I first tried partial clear with 'logical', that too took considerable amonut of time, and the
difference between logical and physical clear is only 15-20 minutes. FYI, I have 31 dimensions in this cube,
and the MDX clear script that use Scenario->ACTUAL and Period->&CLEAR_PERIOD (SubVar) is of dynamic hierarchy
type.
Is there a way I can rewrite the clear data MDX script in betterway  so that it will clear faster, than this
<<CrossJoin({([ACTUAL])},{[&CLEAR_PERIOD]})>>
Does this clear MDX have any effect on dynamic/stored hierarchy nature of the dimension, if not, then what
would be optimized way to write this MDX?
Thanks,
DD