Architecture of Query Execution—Calculating MDX Expressions - Microsoft SQL Server 2008 Analysis Services - page 523

CHAPTER 29

IN THIS CHAPTER

. Query Execution Stages

Architecture of Query
Execution—Calculating
MDX Expressions

. Parsing an MDX Request

. Creation of Calculation Scopes

. Building a Virtual Set

Operation Tree

. Optimizing Multidimensional

Space by Removing Empty
Tuples

A multidimensional database consists of many parts, but
at its heart are two major components: the storage engine
and the formula engine. The storage engine writes data to
and reads data from the disk. The formula engine accepts
Multidimensional Expressions (MDX) requests, executes
them, and sends requests to the storage engine to retrieve
data. The user or client application interacts with the
formula engine, and the formula engine in turn interacts
with the storage engine.

. Calculating Cell Values

. Cache Subsystem

In this chapter, we delve deeper in the discussion how the
formula engine processes MDX query commands, and
examine which data structures and algorithms it uses to
retrieve and calculate necessary results in the shortest possi-
ble time. In Chapter 30, “Architecture of Query Execution—
Retrieving Data from Storage,” we discuss how the storage
engine processes request for data from the formula engine.

Optimization of MDX calculations was one of the most
important goals of Analysis Services 2008. Another goal of
this release was not only to improve the performance of
certain queries but also to make the system's behavior more
consistent and predictable. With those goals in mind,
Analysis Services developers redesigned the calculation
engine, building on top of the successful design approaches
introduced in the preceding version. In this redesign, some
areas that we discuss shortly were changed, and new opti-
mizations were introduced.