Database Reference
In-Depth Information
likes things in a certain order to allow it to perform best. violate that order and things
slow down, as you will see below.
A report in Smart view can be thought of as having four axes: point of view (Pov),
Page, row, and Column. While Pov and Page may seem very similar, and they are,
there is an important distinction. Pov members are single members that are off-grid,
e.g., the member Actual. Page dimension members are also single member selections,
but within a set of members, e.g., Jan is selected in the set of 12 months Jan through
Dec. If more than one member of a dimension is on-grid, it is either a row or Column
dimension. A large number of choices in the Page will negatively affect performance.
For block storage option (BSo) databases, a report or spreadsheet with dense dimen-
sions in both the rows and Columns will retrieve much faster than one with sparse
dimensions on the rows and Columns. The implication of this is, while designing your
reports and worksheets, try to keep the row and Column dimension choices limited to
dense dimensions as much as is practical.
10.3.6 Attributes, DTS, and Dynamically Calculated Members
The impact of attributes and dynamically calculated members on reports are cov-
ered at length in this topic (see Chapter 4, BSo Will never Die and here Is Why,
and Chapter 7, how ASo Works and how to Design for Performance). From a query
perspective, it is important to know that any member that is dynamically calculated
during a retrieve will negatively impact retrieve performance. In the BSo architec-
ture, the main culprits in this performance consideration are Attribute Dimensions,
Dynamically Calculated members, and DtS (Dynamic time Series) members. ASo
doesn't have the block concept, but its dynamic nature means that queries to nonag-
gregated views, formulas, and attribute combinations may be similarly slow. Clearly you
cannot create spreadsheets with only precalculated members, but if you keep in mind
the performance impact these members have, you will be able to manage your design
with an eye on performance.
10.3.7 Use “Navigate Without Data” to Improve Performance during Development
navigate Without Data does exactly what it says, it allows Smart view to retrieve meta-
data without the corresponding data. This can be useful when creating expensive queries.
Perhaps the most helpful aspect of the navigate Without Data option setting is its abil-
ity to greatly enhance performance. When a sheet's members are programmatically
manipulated more than once, such as moving a member from the Pov to the grid
then pivoting the grid, it is a good practice to enable navigate Without Data during the
sheet manipulations. Enabling this setting allows the sheet to refresh several times very
quickly because only metadata is queried.
10.3.8 Compression
network performance of Smart view can be enhanced using compression. Smart view
providers use a compression scheme called gzIP. Because the communication between
the client and the server is primarily alphanumeric text, using gzIP compression greatly
increases efficiency.
to get an idea just how efficient gzIP compression is, simply use your favorite
zip utility to zip a fairly large data load file. In most cases, the compressed file will
be 80% smaller than the nonzipped version. you can expect similar results and
