RADS SEMINAR

 

PLACE AND TIME: Thursday, March 15, 2:30 pm, CB 5111

 

TITLE: “Choice of Aggregation Groups for Layered Performance Model Simplification

Farhana Islam, Dorina Petriu, Murray Woodside”

 

The authors previously showed that a complex layered performance model could be simplified by aggregating the contributions of subsystems, following a few simple principles which give good accuracy in many cases. The question of which subsystems to merge is further examined here, leading to identifying groups of subsystems (corresponding to “tasks” in layered queueing models) which can be safely aggregated. The grouping begins by identifying tasks which should be preserved, not aggregated, including those which are (or might become) bottlenecks. Then the groups are defined by their relationship to these preserved tasks.

 

The talk will explain the aggregation groups and show examples.

 

 

 

RADS SEMINAR

 

PLACE AND TIME: Thursday, March 8, 2:30 pm, CB 5111

TITLE: Automatic transformation of UML+MARTE models to LQN performance models

 

PRESENTERS: Prof. Dorina Petriu and PhD Student Taghreed Altamimi

 

SUMMARY:

In the first part of the talk we discuss how to use specialized model transformation languages, such as Epsilon ETL, for generating LQN performance models from UML+MARTE software models. Specialized languages help create a more compact transformation, easier to understand and maintain than transformations developed with general purpose languages, such as Java. Beside the performance model, the transformation also generates a traceability model containing trace links

between the software and performance model. After solving the performance model, the performance results are fed back to the software model by following in reverse the cross-model trace links. When the UML model evolves during the development process, the traditional solution for keeping

 


RADS SEMINAR

 

PLACE AND TIME: 4539 MacKenzie, Mon Feb 26, 10.30 AM

 

TITLE: Multi-level tracing for large scale manycore heterogeneous distributed systems

 

Presenter: Prof. Michel Dagenais, Polytechnique Montréal

 

SUMMARY:

 

The presentation will briefly introduce the organisation of the DORSAL laboratory and the collaborative research and development projects undertaken with the financial support of industrial partners. Then, several recent results in the area of kernel, GPU and user-level tracing and trace analysis will be outlined, with emphasis on real-time, distributed, multi-core and heterogeneous Linux systems. Different widely used open-source tools developed collaboratively in the DORSAL laboratory, such as LTTng, Trace Compass and the Common Trace Format, will be described, as well as how they were exploited to quickly find the critical path and uncover hard to find problems in several real industrial systems.

 

Bio

 

Michel Dagenais is a professor at Ecole Polytechnique de Montreal in Computer and Software Engineering. He authored or co-authored numerous scientific publications, as well as free documents and free software packages in the fields of operating systems, distributed systems, and multicore systems, particularly in the area of tracing and monitoring Linux systems for performance analysis. He collaborates intensively with several large industrial partners such as Ericsson, Ciena, Google and AMD on complex performance analysis problems. The Linux Trace Toolkit next generation, developed under his supervision, is now used throughout the world and is part of several specialised and general purpose Linux distributions.

 

 

RADS SEMINAR

 

PLACE AND TIME: 4359 ME. Thursday, Jan 25 at 2.30 - 3.30 PM

 

Title: Deriving a LQ Performance Model from Traces by SAME and WebTime

 

Presenter Prof Murray Woodside

 

SUMMARY: The best description of a system comes not from the plans for it, but from evidence gathered from the running code, for example as traces of interactions and operations by system components. This way you are guaranteed not to be misled by out-of-date descriptions. The seminar will describe two tools that tackle this question, from past research in the Lab.