New Challenges in Physical Database Design - Automated Physical Database Design and Tuning - page 225

indexes for a single query. These primitive data structures and algo-
rithms are not necessarily meant to be consumed by DBAs but instead
provide a foundational abstraction for building new applications. Using
these data structures and algorithms, we can rather easily build func-
tionality similar to existing physical design tools and can package it in
yet another monolithic tool. However, a more interesting alternative is
to expose all data structures and algorithms in a library that can be
leveraged and extended by experienced DBAs.
Front Ends: Front ends are based on both the low- and high-level APIs and
deliver functionality to end users. A very powerful interaction model
is a scripting platform to interact with the physical database design.
The scripting language understands the data structures and algorithms
exposed by the underlying layers and allows users to write small inter-
active scripts to tune the physical design of a database. Common tasks,
such as minimizing cost for a single storage constraint, obtaining the
best configuration for a given query, or other functionality provided by
current physical design tools, can be implemented as scripts accessed
using graphical user interfaces by relatively inexperienced DBAs.
12.3.2 A Sample Tuning Session
We next illustrate an interactive tuning session using a prototype built on
top of Windows PowerShell, an interactive scripting language that integrates
with the Microsoft .NET Framework. Windows PowerShell provides an envi-
ronment to perform administrative tasks by execution of cmdlets (which are
basic operations), scripts (which are compositions of cmdlets), or stand-alone
applications or by directly instantiating regular .NET classes. The example
leverages a PowerShell provider that exposes all the information about a tun-
ing session in a hierarchical and intuitive object model (analogous to a file
system). By using this provider, we can navigate and manipulate the state of
a tuning session easily. Additionally, we exploit several scripts that are built
by leveraging the high-level APIs discussed earlier. Figure 12.4 illustrates how
an interactive approach can benefit DBAs by providing flexibility and control
during physical database design. We expect that advanced DBAs could create
their own scripts to further customize the physical design tuning experience.
Interactive sessions have the potential to change the way DBAs think about
physical database design.
12.4 Physical Database Design Benchmarks
Although there has been considerable research in techniques to address the
physical design problem, much less attention has been paid to systematic
methodologies to evaluate the quality of different approaches. Instead, each