Evaluating similarity-based trace reduction techniques for scalable performance analysis

Mohror, Kathryn; Karavanic, Karen L.

doi:10.1145/1654059.1654115

Cited by 28 publications

(15 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Trace reduction is the compression of traces in some manner (either lossless or lossy) so that they can be stored and processed efficiently (Kaplan et al 1999;Mohror and Karavanic 2009). The process of collapsing the commit graph can be seen as a particular case of trace reduction.…”

Section: Trace Reduction Methods and Automatic Collapsingmentioning

confidence: 99%

DyeVC: an approach for monitoring and visualizing distributed repositories

Cesario

Interian

Murta

2017

J Softw Eng Res Dev

View full text Add to dashboard Cite

Software development using distributed version control systems has become more frequent recently. Such systems bring more flexibility, but also greater complexity to manage and monitor multiple existing repositories as well as their myriad of branches. In this paper, we propose DyeVC, an approach to assist developers and repository administrators in identifying dependencies among clones of distributed repositories. It allows understanding what is going on around one's clone and depicting the relationship between existing clones. DyeVC was evaluated over open source projects, showing how they could benefit from having such kind of tool in place. We also ran an observational and a performance evaluation over DyeVC, and the results were promising: it was considered easy to use and fast for most repository history exploration operations while providing the expected answers.Keywords: Distributed version control, Monitoring, Visualization, Awareness BackgroundVersion Control Systems (VCS) date back to the 70s when SCCS emerged (Rochkind 1975). Their primary purpose is to keep software development under control (Estublier 2000). Along these almost 40 years, VCSs have evolved from a centralized repository with local access (e.g., SCCS and RCS (Tichy 1985)) to a client-server architecture (e.g., CVS (Cederqvist 2005) and Subversion (Collins-Sussman et al. 2011)). More recently, distributed VCSs (DVCS) arose (e.g., Git (Chacon 2009) and Mercurial (O'Sullivan 2009a)) allowing clones of the entire repository in different locations. According to a survey conducted by the Eclipse community (2014), Git and GitHub combined usage increased from 6.8 to 42.9% between 2010 and 2014 (a growth greater than 500%). During this same period, Subversion and CVS combined usage decreased from 71 to 34.4%. This clearly shows momentum and a strong tendency in the adoption of DVCSs in the open source community.Besides these changes from local to client-server and then to a distributed architecture, the concurrency control policy adopted by VCSs also changed from lock-based (pessimistic) to branch-based (optimistic). According to Walrad and Strom (Walrad and Strom 2002), creating branches in VCSs is essential to software development because it enables parallel development, allowing the maintenance of different versions of a system, the customization to different platforms/customers, among other features. Cesario et al. Journal of Software Engineering Research and Development (2017) 5:5 DOI 10.1186/s40411-017-0039-8 the branch creation into a recurring pattern, no matter if this creation is explicitly done by executing a "branch" command or implicitly when a repository is cloned.However, distributed software development, especially from the geographical perspective (Gumm 2006), brings a set of risk factors, and Configuration Management (CM) is affected by them. The increasing growth of development teams and their distribution along distant locations, together with the proliferation of branches, introduce additional complexity for ...

show abstract

Section: Trace Reduction Methods and Automatic Collapsingmentioning

confidence: 99%

DyeVC: an approach for monitoring and visualizing distributed repositories

Cesario

Interian

Murta

2017

J Softw Eng Res Dev

View full text Add to dashboard Cite

show abstract

“…Our method is based on an idea by Mohror and Karavanic, who introduce a method that uses wavelets for direct similarity checking using distance metrics [29]. However, we generalise the idea under the assumption that comparing time and frequency coefficients between the complementary wavelet components of the two signals provides a more realistic comparison than comparing their distance directly.…”

Section: Methodsmentioning

confidence: 99%

A similarity measure for time, frequency, and dependencies in large-scale workloads

Lassnig

Fahringer

Garonne

et al. 2011

Proceedings of 2011 International Conference for High Performance Computing, Networking, Storage and Analysis

View full text Add to dashboard Cite

Performance evaluations of large-scale systems require the use of representative workloads with certifiable similar or dissimilar characteristics. To quantify the similarity of the characteristics, we describe a novel measure comprising two efficient methods that are suitable for large-scale workloads. One method uses the discrete wavelet transform to assess the periodic time and frequency characteristics in the workload. The second method evaluates dependencies in descriptive attributes via association rule learning. Both methods are evaluated to find the limits of their similarity spaces. Additionally, the wavelet method is evaluated against existing similarity methods and tested for noise robustness and random bias. An empirical study using workloads from seven operational large-scale systems evaluates the measure's accuracy. The results show that our measure is highly resistant to noise, well-suited for large-scale workloads, covers 87% of the possible similarity space, and improves accuracy by 24.5% and standard deviation by 10.8% when compared to existing work.

show abstract

“…Low memory per core ratios require an efficient management of trace data, which has been previously addressed by OTF2 [14], cCCG [15], a study of reduction techniques [16], and ScalaTrace [17]. While the latter three approaches are capable of reducing the trace data to a nearly constant trace size (depending on the granularity of the aggregation) they may be very time consuming 1 .…”

Section: Related Workmentioning

confidence: 99%

“…In the section thereafter, we describe our event data compression and reduction management within a fixed-size memory footprint. 1 The approaches in [16] and [17] do not give results for runtime overheads.…”

Section: Infrastructurementioning

confidence: 99%

Online Performance Analysis: An Event-Based Workflow Design towards Exascale

Wagner

Hilbrich

Brunst

2014

2014 IEEE Intl Conf on High Performance Computing and Communications, 2014 IEEE 6th Intl Symp on Cyberspace Safety and Security

View full text Add to dashboard Cite

Today, it is commonly accepted that speedup and efficiency are not granted automatically when developing or porting software for High Performance Computing (HPC) platforms. The reasons are manifold and actively investigated in the community. Software monitors are essential for these studies as they provide the raw performance data to be analyzed. We propose an online monitoring workflow for event-based performance analysis that takes into account the significant changes in system architecture towards the development of exascale supercomputers. Critical properties are: communication across large numbers of processing elements, limited I/O capabilities, and a decreasing memoryper-core ratio. We present a hierarchical data management and steering workflow that directly couples the application, measurement, and analysis processes, thus eliminating the need for extensive communication and buffering. The workflow is closely integrated with the native system communication API to enable best communication across processing elements. The memory issue is addressed with a new lossy hierarchical data compression technique for in-memory storage, intended for small, fixed-size buffers. Further, we abandon secondary storage to avoid potential I/O challenges. We demonstrate the feasibility of our design with a prototype implementation that features services for data collection, analysis, and runtime compression. Our evaluation extrapolates results obtained with the NAS Parallel Benchmarks at up to 2,048 processes to an exascale workflow.978-1-4799-6123-8/14 $31.00

show abstract

Evaluating similarity-based trace reduction techniques for scalable performance analysis

Cited by 28 publications

References 41 publications

DyeVC: an approach for monitoring and visualizing distributed repositories

DyeVC: an approach for monitoring and visualizing distributed repositories

A similarity measure for time, frequency, and dependencies in large-scale workloads

Online Performance Analysis: An Event-Based Workflow Design towards Exascale

Contact Info

Product

Resources

About