Hai sir any body please answer to the query to which i am posting (please)

Question

hi
any body please to the query which iam posting. i could not what exactly happens i was thinking so much even then i could not get the answer.
Q) When we use the fix on dense or sparse on which dimension performance is good? How can you justify that?
(IN PDF it is written this way)
When you use the FIX command only on a dense dimension, Analytic Services retrieves the
Entire block that contains the required value or values for the member or members that you
Specify. Thus, I/O is not affected, and the calculation performance time is improved.
When you use the FIX command on a sparse dimension, Analytic Services retrieves the block
for the specified sparse dimension member or members. Thus, I/O may be greatly reduced.
I cannot justify (visualize the above answer can any body help me the way it is going to happen from disk to the memory in the point of the performance) please elaborate what exact process takes.

Answer 1

Hi,
In Essbase (aka Analytic Services) block storage option (BSO) databases, the data block is the basic unit of I/O. A block contains cells that represent the intersections of stored members from dense dimensions. Unique data blocks are created for the existence of sparse dimension combinations (based on data loads or calculations).
In a calc script, a FIX on a dense member will require the calculator to perform I/O on ALL the existing data blocks in the database. This may be logically necessary in some cases but it is not efficient in terms of minimizing I/O. Worse yet, a series of dense FIX statements (not nested but one after the other) will cause multiple, full passes on all data blocks. Conversely, a FIX on a sparse member will only require I/O on the data blocks related to that sparse member via the index. This is where you get I/O efficiency.
To illustrate this in a simplistic example, let's say you have 1 million data blocks. One of the dense dimensions is Period with twelve months rolling up to quarters and total year. One of the sparse dimensions is Scenario with ten distinct scenarios (Actual, Budget, Forecast, etc.) and a Label Only setting at the top of the dimension (i.e., on Scenario at Gen 1). Consider the I/O differences on each FIX example to follow...
FIX(Jan)
... any calc action here must perform I/O on all 1 million blocks
ENDFIX
FIX(Feb)
... any calc action here must perform I/O on all 1 million blocks AGAIN if written in sequence like this after the Jan FIX above
ENDFIX
However...
FIX(Budget)
... any calc action here must perform I/O on only 1/10th of the database or 100,000 of the blocks
ENDFIX
Now these examples oversimplify some things to make the basic point about which you have asked a question. I hope this helps. Please post any follow-up questions.
Good luck, Darrell Barr