During analysis for Cache memory I accidentally found few new features of BW 7.3 (EHP 1)
*(Please ignore if you already know this or not new 
)
Generally for process chain data load, we use delete index and create index option. This will make the data load fast and also aviod the locking error becasue of the indexes.
Similarly one should update the statistics data as well roll up data in cube (for aggregates).
Why should I maintained it in PC level, if I am loading the data into same infoProvider (say Cube) using multiple PCs.
Why not at Cube level.
Answer is with BW 7.3 via T Code RSDIPROP
As description indicate
In this transaction you edit the performance, data load, and
query properties of an InfoProvider
Go to DB Performance and get various options to manage the data load as below
The F1 help available for more info.
Also Other tab page like Roll Up will give option to select the roll up process used for aggregate and Cache setting options at cube level to improve report performance.
DSO and even infoobject properties can also be set if they are also infoproviders
I found it useful and hence to share.
Thank-You.
Regards,
VB