Dynamic Analysis and Profiling of Multithreaded Systems - Advanced Operating Systems and Kernel Applications - page 194

ing is particularly useful where fine-grained event
data can be collected and used to derive charac-
teristics of a running system. Dynamic analysis
is weaker and less capable, when the behavioral
characteristic depends on system-wide analysis,
such as the global thread state. It is therefore
clear that runtime profiling alone is insufficient
to capture and predict a complete image of system
behavior due to the “as observed” syndrome, that
is, dynamic analysis can only assure statistical
certainty of behavior because it just collects be-
havioral data for a given execution trace.
The alternative to dynamic analysis is static
analysis, such as program analysis and model
checking. The benefits of static analysis are its
ability to (1) perform analysis without running
the system (useful for pre-integration testing),
and (2) allow the inspection of all theoretically
possible (albeit less frequent) conditions. Although
static analysis is promising in some areas, it also
cannot capture and predict a complete image of
behavior for large-scale systems. In particular,
static-analysis techniques are limited in their
practical applicability (e.g., scalability) and in
their ability to relate to wall-clock time.
Behavioral analysis technology will be increas-
ingly important as the systems we build become
larger, more parallel, and more unpredictable.
New tools and techniques that strategically
combine static and dynamic analysis—and that
partition the system into well-defined “behavioral
containers”—will be critical to the progression
along this path.

Artho, C., & Biere, A. (2005). Combined static
and dynamic analysis. In Proceedings of the 1 st
International Workshop on Abstract Interpreta-
tion of Object-oriented Language (AIOOL 2005) ,
ENTCS, Paris. Elsevier Science Publishing.

Baxter, I. (2004). DMS: Program transformations
for practical scalable software evolution. In Pro-
ceedings of the 26 th International Conference on
Software Engineering , (pp. 625-634).

Beck, M., Bohme, H., Dziadzka, M., Kunitz, U.,
Magnus, R., & Verworner, D. (1999). Linux Kernel
internals (2nd ed.) . Addison Wesley Longman.

Binder, W. (2005). A portable and customizable
profiling framework for Java based on bytecode
instruction counting. In Proceedings of the Third
Asian Symposium on Programming Languages
and Systems (APLAS 2005) , (LNCS 3780, pp.
178 -194).

Binder, W., & Hulaas, J. (2004, October). A
portable CPU-management framework for Java.
IEEE Internet Computing , 8 (5), 74-83.

Binder, W., Hulaas J., &Villaz A. (2001). Por-
table resource control in Java. In Proceedings of
the 2001 ACM SIGPLAN Conference on Object
Oriented Programming, Systems, Languages and
Applications , (Vol. 36, No. 11, pp. 139-155).

Boner, J. (2004, March). AspectWerkz—Dynamic
AOP for Java. In Proceedings of the 3 rd Interna-
tional Conference on Aspect-oriented develop-
ment (AOSD 2004). Lancaster, UK.

Borland Software Corporation. (2006). Borland
Optimize-it Enterprise Suite (Computer software).
Retrieved March 11, 2008, from http://www.bor-
land.com/us/products/optimizeit/index.html

referenceS

Arnold, M., & Ryder, B. G. (2001). A framework
for reducing the cost of instrumented code. In
Proceedings of the SIGPLAN Conference on
Programming Language Design and Implementa-
tion , (pp. 168-179).

Broberg, M., Lundberg, L., & Grahn, H. (1999,
April). Visualization and performance predic-
tion of multithreaded solaris programs by trac-
ing kernel threads. In Proceedings of the 13th
International Parallel Processing Symposium ,
(pp. 407-413).