Wednesday, 13 December 2017

Poster Session

Phymss - Performance Hybrid Model Solver and Simulator Based on UML MARTE Diagrams

Authors:

Cosmina Chişe (University of Timisoara)
Ioan Jurca (University of Timisoara)

Abstract:

There are several research directions in Software Performance Engineering (SPE), covering the entire performance prediction process, but most of the tools developed so far implement only part of it or have restrictions. From a methodology perspective, current performance prediction tools rely either on analytical or simulation models, as separate techniques. This paper presents a performance analysis tool, Phymss (Performace Hybrid Model Solver and Simulator), which covers the analysis process from the input system model annotated with performance information to obtaining performance results and inserting them back into the original system model. Two analysis methods are implemented, for flexibility reasons: a multithreaded simulator and a hybrid solver that combines the analytical and simulation approaches in a new analysis technique, in order to investigate the benefits of such an approach.

DOI: 10.1145/1712605.1712643

Full text: PDF

[#][]

Towards the Identification of "Guilty" Performance Antipatterns

Authors:

Vittorio Cortellessa (Università degli Studi dell-Aquila)
Anne Martens (Karlsruhe Institute of Technology)
Ralf Reussner (Karlsruhe Institute of Technology)
Catia Trubiani (Università degli Studi dell'Aquila)

Abstract:

The problem of interpreting the results of software performance analysis is very critical. Software developers expect feedback in terms of architectural design alternatives (e.g., re-deploy a component), whereas the results of performance analysis are pure numbers. Support to the interpretation of such results that helps to fill the gap between numbers and software alternatives is still lacking. Performance antipatterns can play a key role in the search of performance problems and in the formulation of their solutions. In this poster, we introduce a process to elaborate the analysis results and to score performance requirements, model entities and "guilty" performance antipatterns.

DOI: 10.1145/1712605.1712644

Full text: PDF

[#][]

On the Efficacy of Call Graph-Level Thread-Level Speculation

Authors:

Arun Kejariwal (Yahoo! Inc.)
Milind Girkar (Intel Corporation)
Xinmin Tian (Intel Corporation)
Hideki Saito (Intel Corporation)
Alexandru Nicolau (University of California at Irvine)
Alexander V. Veidenbaum (University of California at Irvine)
Utpal Banerjee (University of California at Irvine)
Constantine D. Polychronopoulos (University of Illinois at Urbana-Champaign)

Abstract:

Thread-level speculation (TLS) has been proposed as a means to parallelize difficult-to-analyze sequential codes. In this paper, we present a realistic measure of the performance potential of call-graph level TLS, using the SPEC CPU2006 benchmark suite and the Intel Core 2 Duo processor.

DOI: 10.1145/1712605.1712645

Full text: PDF

[#][]

Semantic Layered Architecture to Integrate FR/NFR in Software Performance Engineering

Authors:

Isaac Lera (University of the Balearic Islands)
Ramon Puigjaner (University of the Balearic Islands)

Abstract:

The integration process using UML Profile for Schedulability, Performance, and Time Specification (SPT) or MARTE specification presents some deficiencies due to the inability to formalize the representation criteria and to relate information in a transparent way between the outputs of performance models and software designs. The drawbacks of FR and NFR model integration appears to be crucial for the real application of several SPE techniques, since the performance modelling and the software modelling are not semantically related because there are not explicit and formal semantic interpretation of the concepts. To avoid this, we propose a semantic architecture based on well-known Web Semantic Layered Architecture.

DOI: 10.1145/1712605.1712646

Full text: PDF

[#][]

Automatable & Scalable Late Cycle Performance Analysis

Authors:

Florian Mangold (Universität München)
Moritz Hammer (Universität München)
Harald Roelle (Siemens AG)

Abstract:

Performance analysis of large, concurrent systems is a difficult problem that can hardly be approached with classical profiling. Performance issues might be caused by the interaction of modules and hardware components, making it difficult to find exact causes by considering single modules. By slowing down single modules artificially, dependencies of modules can be detected. Employing statistical means, such dependencies are detected in the covariance of runtime changes. We propose a way to detect the most meaningful dependencies in large-scale systems, allowing arbitrary scaling with respect to the granularity considered.

DOI: 10.1145/1712605.1712647

Full text: PDF

[#][]

Assessing Identity and Access Management Systems Based on Domain-Specific Performance Evaluation

Authors:

Frank Schell (Karlsruhe Institute of Technology)
Andreas Schaf (Karlsruhe Institute of Technology)
Jochen Dinger (Karlsruhe Institute of Technology)
Hannes Hartenstein (Karlsruhe Institute of Technology)

Abstract:

Identity and access management (IAM) systems are used to assure authorized access to services in distributed environments. The design decisions of IAM systems, in particular the arrangement of the involved components, have significant impact on performance, access control accuracy, and costs of the overall system. Hence, systematic engineering of IAM systems demands for criteria and metrics to differentiate architectural approaches. Therefore, we propose a system design framework for IAM that can be used to evaluate different design decisions in advance.

DOI: 10.1145/1712605.1712648

Full text: PDF

[#][]

PMIF Extensions: Increasing the Scope of Supported Models

Authors:

Connie U. Smith (Performance Engineering Services)
Catalina M. Lladó (Universitat de les Illes Balears)
Ramon Puigjaner (Universitat de les Illes Balears)

Abstract:

Performance model interchange formats are common representations for data that can be used to move models among modeling tools. In order to manage the research scope, the initial version of PMIF is limited to QNM that can be solved by efficient, exact solution algorithms. The overall model interoperability approach has now been demonstrated to be viable. This paper is a first step to broaden the scope of PMIF to represent models that can be solved with additional methods.

DOI: 10.1145/1712605.1712649

Full text: PDF

[#][]