A new algorithm in the Aggregation Designer
helps you create initial aggregations. This designer is
optimized to work with usage-driven aggregations.
You can view the created aggregations and add to or
remove them.
Dynamic named sets are a new capability of SSAS
2008. A named set in SQL Server 2005 makes it possible
to define a set of dimension members such as a
set of the top 10 stores by sales. You define this set
statically. You can then refer to this named set wherever
you need to see the top 10 stores by sales. In SQL
Server 2005, set evaluation occurs only at set creation.
In SQL Server 2008, you can create dynamic named
sets and define them to be evaluated every time the
sets are used.
Performance enhancements. A major portion of
the SSAS performance enhancements are in areas
such as subspace computations, Multidimensional
OLAP (MOLAP)-enabled write-back, and backup
and storage.
Cube space is generally sparse, with values existing
only for a small number of dimension intersections.
Although SSAS in SQL Server 2005 evaluates
expressions on complete space, and subspace computation
is included with SP2, in SQL Server 2008 SSAS
subspace computations are significantly improved.
Multidimensional Expressions query performance
has improved; SSAS deals better with cube space by
dividing the space to separate calculated members,
regular members, and empty space to improve evaluation
of cells that need to be included in calculations.
The new MOLAP-enabled writeback capabilities
in SQL Server 2008 SSAS remove the need to
store writeback data in ROLAP storage mode. The
new writeback MOLAP storage mode results in significant
performance gains in cubes that leverage the
writeback capabilities.
Finally with SQL Server 2008 SSAS backup
compression, less storage is required to keep backups
online. The backups also run significantly faster
because less disk I/O is required. There are fewer
restrictions on the size of the database, and the
time required for backup and restore operations is
significantly reduced.
Data mining. The next important SSAS element
for any BI solution is data mining. SQL Server 2008
SSAS enhances data mining models by appending
a new algorithm to the Microsoft Time Series algorithm.
This improves the accuracy and stability of
predictions in the data mining models. The new algorithm
is based on the Auto-Regressive Integrated
Moving Average (ARIMA) algorithm, and provides
better long-term predictions than the Auto Regression
Trees with Cross Predict (ARTxp) algorithm
used in SQL Server 2005 SSAS.
By default, the new implementation of the Microsoft
Time Series algorithm uses the ARTxp algorithm
to train one version of the data mining model and the
ARIMA algorithm to train another version of the
data mining model. The algorithm then weighs the results
of these two data mining models to provide the
prediction characteristics you want. If you don’t want
to use the default implementation, you can specify the
algorithms that the Microsoft Time Series algorithm
must use.
In SQL Server 2008 Enterprise Edition, you can
specify a custom weighting of the algorithms to provide
the best prediction over a variable time span.
The improved Microsoft Time Series algorithm accepts
data during prediction to allow for new business
scenarios. For example, you can create a revenue
prediction model based on averages across products,
regional aggregates, or some other broad data set.
You can then apply that model to the time series that
shows the sales of an individual product. By applying
the general model, you can take advantage of the stability
and availability of aggregate data and customize
prediction to the individual product. You can also
train models by using multiple series, and then apply
the models to new data in forecasting scenarios.
SSRS Enhancements
Now that we’ve covered what’s new with laying the BI
groundwork with SSIS and building cubes and mining
models in SSAS, we’re ready to review the new
features and enhancements found in SSRS in SQL
Server 2008.
Report Server engine. A report server is now
implemented as a Windows-based service that hosts
the Report Manager, the Report Server Web service,
and background processing feature areas. The report
engine improves supportability and the ability to control
server behavior with memory management and
infrastructure consolidation. Consolidating server
applications into a single service reduces configuration
and maintenance tasks. However, the Report
Manager and the Report Server Web service applications
continue to run independently within the single
service. Both the Report Manager and the Report
Server Web service can be accessed through URLs
that provide HTTP access to these applications.
The report server includes an HTTP listener that
handles all authentication requests directed to a URL and a port you define during server configuration. To
provide the ASP.NET and Report Server Web service,
the report server uses the new HTTP.SYS capabilities
of the OS instead of IIS. The report server also has
new management features to set a memory threshold
for background operations and performance counters
for monitoring service activity. I’ll briefly explore
SSRS enhancements for report server deployment
modes, report authoring, and report designing.
SSRS continues to expand its delivery options
with the expansion and enhancement of Rich Text
Format (RTF), Microsoft Office Word, and Microsoft
Office Excel rendering. The improvement of
the RTF component provides a method for users
to define mixed formatting in textboxes and import
marked-up strings of the text into a report generated
from a database or other data sources. The Microsoft
Office Word 2007 rendering extension can be used to
export a report to a Word document without using a
third-party tool. Finally, the Microsoft Office Excel
rendering extension has been enhanced to support
features such as nested data regions and sub-reports.
Continued on page 3
Prev. page
1
[2]
3
next page 
