Database Reference
In-Depth Information
Smart view. The Smart view developers were keen to recognize the importance of this
and the synergy that results.
Like the speedometer in your car, both Excel and Smart view “advertise” an optimistic
row limit capability. This number is merely a concept and, again like your car, neither is
really capable of functioning with that many rows nor should you try. Excel and Smart
view both prefer data that is organized into a smaller number of columns and a larger
number of rows. typically this means periods and maybe a couple of scenarios in the
columns, the rest in the rows. Shortly, we will explore the tuning implications of these
properties and more. This chapter assumes you are experienced with Excel and have
an understanding of efficient use of Excel's formula and calculation capabilities. If not,
consider brushing up before proceeding.
Because much of the rendering done by Smart view happens in Excel, much of the
performance tuning can be done here as well. This is great news because now you can
take performance into your own hands instead of calling the network engineer and
hearing “it's not my fault; the networks look fine from here.”
10.3.2 How Smart View Is Architected; What Are All Those URLs?
Smart view is a middleman service that operates in a client-server fashion. The Smart
view server is a Web service that passes information back and forth from itself to Excel
(your laptop) via Web service calls, which in simple terms are xmL strings. This means
that each of those urLs you configured in Smart view is a Web service urL.
The Smart view server, which is called Analytic Provider Services (APS), is the
server that runs the Web service. APS communicates with the Essbase server in an
Essbase-native format using the Essbase Java API. The server then converts the results
from Essbase into xmL and passes the xmL back to the waiting Smart view client.
The Smart view client is responsible for reading the xmL returned from the server and
rendering the results in Excel as your data and metadata.
10.3.3 Spreadsheet Efficiency Meets Essbase
The implication of this is that not only is your hardware important, but how you con-
struct your reports and worksheets is equally if not more important for maximum
retrieval speed. two key components come into play here: (1) the size of the data set
(xmL) being requested from, or sent to, the Smart view server, and (2) the configura-
tion of the query that Smart view uses against Essbase. In other words, if you create a
huge spreadsheet that has an inefficient arrangement of dimensions, the resulting xmL
string passed over the network and parsed at each end will be huge and the Essbase
query executed on the server will run very slowly. By creating smaller, efficient sheets,
you can dramatically increase the performance of Smart view all by yourself.
10.3.4 How Many Cells?
Sometimes people like to create a report that is 17 columns (months, quarters, total
year) or more across and with many rows. If this report has 200 rows, then it also has
3400 cells. While that is not a huge number to Essbase, it is significant. Limit the num-
ber of cells and you are almost guaranteed to have snappy spreadsheets.
10.3.5 Dimension Arrangement for Efficient Spreadsheets
Efficient spreadsheets are a simple matter if you follow a few rules. The reason for this is
the layout of the sheet is reflected by the construction of the query to Essbase. Essbase
