Session 7: Poster Preview Presentations

Apache Hadoop Performance-Tuning Methodologies and Best Practices

Authors:

Shrinivas B. Joshi (Advanced Micro Devices, Inc.)

Abstract:

Apache Hadoop is a Java based distributed computing framework built for applications implemented using MapReduce programming model. In recent years, Hadoop technology has experienced an unprecedented growth in its adoption. From single-node clusters to clusters with well over thousands of nodes, Hadoop technology is being used to perform myriad of functions - search optimizations, data mining, click stream analytics, machine learning to name a few. Although setting up Hadoop clusters and building applications for Hadoop is a well understood area, tuning Hadoop clusters for optimal performance is still a black art. In this demo paper, we will attempt to provide the audience with a holistic approach of Hadoop performance tuning methodologies and best practices. We discuss hardware as well as software tuning techniques including BIOS, OS, JVM and Hadoop configuration parameters tuning.

Full text: PDF

[#][↑]

Find Your Best Match: Predicting Performance of Consolidated Workloads

Authors:

Danilo Ansaloni (University of Lugano)
Lydia Y. Chen (IBM Research Zürich Laboratory)
Evgenia Smirni (College of William and Mary)
Akira Yokokawa (University of Lugano)
Walter Binder (University of Lugano)

Abstract:

Modern multicore platforms allow system administrators to reduce the costs of the IT infrastructure by consolidating heterogeneous workloads on the same physical machine. To this end, it is important to develop efficient profiling techniques and accurate performance predictions to avoid violating service-level objectives. In this work we present Tresa, a novel tool to automatically characterize workloads and accurately estimate the execution time of different consolidations. These results can be used to optimize consolidations depending on service-level objectives.

Full text: PDF

[#][↑]

Importing PMIF Models into PIPE2 using M2M Transformation

Authors:

Pere Bonet (Universitat de les Illes Balears)
Catalina M. Lladó (Universitat de les Illes Balears)

Abstract:

Model-to-model (M2M) transformation is a key aspect of model-driven development (MDD), where importing and exporting models fits very well. A queueing network based metamodel (PMIF) and a Petri net metamodel are specified using the Eclipse Modelling Framework. The transformation from PMIF models to Petri net models is then build using ATL. This paper presents such a transformation and its integration into PIPE2, a Petri net modelling tool. It also illustrates the transformation by a simple example.

Full text: PDF

[#][↑]

Kieker: A Framework for Application Performance Monitoring and Dynamic Software Analysis

Authors:

André van Hoorn (Christian-Albrechts-University of Kiel)
Jan Waller (Christian-Albrechts-University of Kiel)
Wilhelm Hasselbring (Christian-Albrechts-University of Kiel)

Abstract:

Kieker is an extensible framework for monitoring and analyzing the runtime behavior of concurrent or distributed software systems. It provides measurement probes for application performance monitoring and control-flow tracing. Analysis plugins extract and visualize architectural models, augmented by quantitative observations. Configurable readers and writers allow Kieker to be used for online and offline analysis. This paper reviews the Kieker framework focusing on its features, its provided extension points for custom components, as well the imposed monitoring overhead.

Full text: PDF

[#][↑]

SPEC - Driving Better Benchmarks

Authors:

Walter Bays (Oracle Corporation)
Klaus-Dieter Lange (Hewlett-Packard Company)

Abstract:

The driving philosophy for the Standard Performance Evaluation Corporation (SPEC) is to ensure that the marketplace has a fair and useful set of metrics to differentiate systems is by providing standardized benchmark suites. This paper gives an overview of SPEC and its continuous drive for better benchmarks.

Full text: PDF

[#][↑]

SPECvirt_sc2010 - Driving Virtualization Innovation

Authors:

David L. Schmidt (Hewlett-Packard Company)
Andrew Bond (Red Hat, Inc.)
Lisa Roderick (VMware Inc.)
Klaus-Dieter Lange (Hewlett-Packard Company)

Abstract:

Overview and future outlook of the #1 industry standard virtualization benchmark, SPECvirt_sc2010.

Full text: PDF

[#][↑]

SPECpower_ssj2008 - Driving Server Energy Efficiency

Authors:

Mike G. Tricker (Microsoft Corporation)
Klaus-Dieter Lange (Hewlett-Packard Company)
Jeremy A. Arnold (IBM Corporation)
Hansfried Block (Fujitsu Technology Solutions GmbH)
Sanjay Sharma (Intel Corporation)

Abstract:

SPECpower_ssj2008 [1] is the first industry-standard SPEC [2] benchmark that evaluates the power and performance characteristics of volume server-class and multi-node class computers. This poster-paper gives an overview of the benchmark that defines the server power measurement standards [8] in the same way SPEC have done for performance.

Full text: PDF

[#][↑]

Server Efficiency Rating Tool (SERT)

Authors:

Mike G. Tricker (Microsoft Corporation)
Klaus-Dieter Lange (Hewlett-Packard Company)
Jeremy A. Arnold (IBM Corporation)
Hansfried Block (Fujitsu Technology Solutions GmbH)
Sanjay Sharma (Intel Corporation)

Abstract:

The Server Efficiency Rating Tool (SERT) [1] has been developed by Standard Performance Evaluation Corporation (SPEC) [2] at the request of the US Environmental Protection Agency (EPA) [3], prompted by concerns that US datacenters consumed almost 3% of all energy in 2010. This poster-paper gives an overview of the SERT

Full text: PDF

[#][↑]

SPEC - Enabling Efficiency Measurement

Authors:

Karl Huppler (IBM Corporation)
Klaus-Dieter Lange (Hewlett-Packard Company)
John Beckett (Dell, Inc.)

Abstract:

An overview of the SPEC PTDaemon and SPEC's Benchmark Methodology.

Full text: PDF

[#][↑]