First International Workshop on Software and Performance ’98

WOSP '98

Santa Fe, New Mexico

October 12-16, 1998

Sponsored by ACM SIGMETRICS, and in cooperation with ACM SIGSOFT, the Computer Measurement Group (CMG), IFIP WG 6.3 and 7.3.

Industrial Sponsors include: Hewlett-Packard Labs, NORTEL, Scientific and Engineering Software (SES).

Conference registration: http://www.sce.carleton.ca/wosp98/register.html

Hotel: http://www.sce.carleton.ca/wosp98/la-fonda.html

Santa Fe: http://www.santafe.org/

Enjoy the cool weather and the old-world, peaceful charm of Santa Fe in October. Author Richard Mahler writes, "Santa Fe was founded four years before Shakespeare wrote The Tempest, 13 years before the Pilgrims landed at Plymouth Rock, and more than 200 years before Napolean’s defeat at Waterloo. Yet Santa Fe has been a center of civilization for even longer than that. Archaeologists have determined that Native Americans lived along the Santa Fe River for centuries before the Spanish ex pedition crossed its banks in 1598."

Santa Fe, NM is in the foothills of the Sangre de Cristo Mountains. It is 7000 ft. above sea level with nearby peaks of 12,000 ft. and it averages 300 days of sunshine. Weather in October is cool – average high temperatures are in the 60s (16-19C) with lows 30-40 degrees (0-5C).

Santa Fe is known for its excellent restaurants for both lunch and dinner. October 10-12 offers a chamber music festival at the Chamber Orchestra Loretto Chapel. Other events in Santa Fe include dinner and dance as well as theater. Please see the October events section on the Santa Fe web page for up to the minute information: http://www.santafe.org/. The Albuquerque Balloon Fiesta takes place the previous week for those who plan an extended visit.

You reach Santa Fe from Albuquerque’s airport (ABQ) by rental car or Shuttlejack bus (505) 982-4311. You can also fly into Santa Fe on United Express (small) commuter flights. The Workshop will be held at the La Fonda Hotel, 100 E. San Francisco St. Space is limited so make reservations early (before September 5) and indicate that you are with the WOSP to get the special conference rate ($138US).

Track 1: Overview Tutorials8:30-10:15 Performance Engineering of SDL/MSC Systems, A. Mitschele-Thiel, Universitat Erlangen, and Bruno Muller-Clostermann, Universitat-GH Essen.

10:45-12:30 Software Performance Engineering and Simulation, R. Puigjaner and N. Vazquez, University of the Balearic Islands, Spain.

1:30-3:15 Design Patterns and Performance, D. Petriu, Carleton University, Canada.

3:45-5:30 Avoiding the Software Performance Crisis, M. Hesselgrave, Lucent Technologies

8:30-12:30 Unified Modeling Language (UML) and Software Performance Engineering, R. Pooley, University of Edinburgh.

1:30-5:30 Performance Engineering Evaluation of CORBA-based Distributed Systems, C. Smith, Performance Engineering Systems, and L. Williams, Software Engineering Research.

Presentation duration (including brief questions): * 10 min, ** 15 min, *** 20 min, followed by a general discussion and question period in each session.

1. 8.30 - 9.15 Workshop Welcome

ISSUE A (Integration): How to Achieve Integration of Methodology for Software Development and Performance Engineering

ISSUE B (Practice): Improving the State of Practice in Software Performance Engineering

ISSUE C (Research): Research Priorities in the Interface between Software and Performance

2. 9.15 - 10.30 Papers: Case Studies

Applying Performance Modelling to a Telecommunication System (**) Christiane Shousha, Dorina Petriu (Carleton University, Canada), Anant Jalnapurkar, Kennedy Ngo (NORTEL, Canada).

Overview of 5ESS-2000 Switch Performance (**), Richard Singer (Lucent Technologies, USA).

Further Experiences with Software Performance Modelling (*) Pete Utton and Gino Martin (British Telecom, UK).

Performance Evaluation of a Knowledge Discovery System (*), A. Inkeri Verkamo (University of Helsinki, Finland).

25 minute panel

3. 11.00 - 12.00 Papers: Methodology 1

POEMS: End-to-end Performance Design of Large Parallel Adaptive Computational Systems (***), Ewa Deelman, Aditya Dube, Adolfy Hoisie, Yong Luo, Richard Oliver, David Sundaram-Stukel, Harvey Wasserman, Vikram S. Adve, Rajive Bagrodia, James C. Browne, Elias Houstis, Olaf Lubeck, John Rice, Patricia Teller, Mary K. Vernon (University of Wisconsin, USA), (AND many more affiliations).

A Wideband Approach to Integrating Performance Prediction into a Software Design Environment (***), Murray Woodside, Curtis Hrischuk (Carleton University, Canada), Bran Selic, Stefan Bayarov (ObjecTime, Canada).

Software Performance Engineering a Digital Signal Processing Application (**), David P. Kelly and Robert S. Oshana (Raytheon Systems Company, USA).

5 minute panel, remainder after Methodology 2

4. 1.30 - 2.30 Papers: Methodology 2

Performance Analysis of Communication Systems Formally Specified in SDL (***), Martin Steppler (Aachen University of Technology, Germany).

On a Language Based Method for Software Performance Engineering of Client/Server Systems (**), Daniel A. Menasce, Hassan Gomaa, (George Mason University, USA).

Modeling Execution Architecture of Software Systems by Colored Petri Nets (**), Jianli Xu, Juha Kuusela, (Nokia Research Center, Finland).

Issues in Cache Management for Commercial Software Systems (*), Prabuddha Biswas (Oracle Corporation, USA).

60 minute panel on Methodology 1 and 2

5. 4.00 - 5.30 ISSUE A (Integration), Session 1:

Integration of Methodology for Software Development and Performance Engineering, Workshop Session on Recommendations (first of three on Issue A: Identify problems and opportunities, open discussion).

6. 8.30 - 10.00 Papers: Practices in Software and Performance

Avoiding the Software Performance Crisis (*), Mary R. Hesselgrave (Lucent Technologies, USA).

Performance Testing of Software Systems (**), Filippos I. Vokolos and Elaine J. Weyuker (AT&T Labs, USA).

Experience in Performance Analysis of Large Real-Time Systems (**), Vesa Hirvisalo (Helsinki University of Technology, Finland).

Applications Optimization Methodology- An Approach (**), Deb Manhardt (Banc One Services Corp, USA).

20 minute panel

7. 10.30 - 11.45 ISSUE B (Practice), Session 1:

Workshop Session on Recommendations (First of three sessions on Issue B: open the discussion)

8. 1.30 - 3.00 Papers: Performance Models of Software

Performance Modeling of Layered Networking Protocol Software Implemented with UNIX STREAMS Facilities: Application to a Frame Relay Network Access Device (**), Adrian E. Conway (GTE Internetworking, USA).

A Multi-Layer Client-Server Queueing Network Model with Synchronous and A synchronous Messages (***), S. Ramesh and H. G. Perros (North Carolina State University, USA).

Performance of Multi-level Client-Server Systems with Parallel Service Operations (***), Greg Franks, Murray Woodside (Carleton University, Canada).

30 minute panel

9. 3.30 - 5.00 Papers: Methodology 3

Managing Performance Requirements for Information Systems (***), Brian A. Nixon (University of Toronto, Canada).

Predicting Memory Use from a Class Diagram using Dynamic Information (**), Gail C. Murphy and Ekaterina Saenko (University of British Columbia, Canada).

Development and Validation of a Hierarchical Memory Model Incorporating CPU- and Memory-Operation Overlap (***), Yong Luo, Olaf M. Lubeck, Harvey Wasserman, Federico Bassetti, and Kirk W. Cameron (Los Alamos National Lab, USA).

20 minute panel

10. 5.00 - 6.00 ISSUE A, Session 2

Workshop Session on Recommendations (second of three sessions on Issue A: formulate recommendations to be discussed on Friday)

11. 8.30 - 10.00 Papers: Architecture and Performance

Performance Evaluation of Software Architectures (***), Lloyd G. Williams (Software Engineering Research, Boulder) and Connie U. Smith (Performance Engineering Services, Santa Fe).

An Approach to Performance Evaluation of Software Architectures (***), S. Balsamo (Univ. di Udine, Italy), P. Inverardi (Univ. dell'Aqila, Italy), C. Mangano (Univ. di Pisa, Italy).

Performance-Oriented Software Architecture Analysis: an Experience Report (**), Chung-Horng Lung, Anant Jalnapurkar, Asham El-Rayess (NORTEL, Canada).

Characterising the Performance of Three Architectural Styles (*), Jan Bosch and Hakan Grahn (University of Karalskrona Ronneby, Sweeden).

20 minute panel

12. 10.30 - 11.45 ISSUE B (Practice), session 2

Improving the State of Practice in Software Performance Engineering. Workshop Session on Recommendations (second of three sessions on Issue B: formulate recommendations to be discussed on Friday).

13. 1.30 - 3.00 Papers: Performance and Workload Measurement

Capacity planning with phased workloads (***), E.Borowsky, R.Golding, P.Jacobson, A.Merchant, L.Schreier, M.Spasojevic, and J.Wilkes (Hewlett Packard Labs, USA).

Exploiting Software Interfaces for Performance Measurement (***), Douglas P. Konkin (PMC-Sierra, Inc.), Gregory M. Oster, and Richard B. Bunt (University of Saskatchewan, Canada).

Correlating Resource Demand Information with ARM Data for Application Services (***) J. Rolia (Carleton University, Canada) and V. Vetland (Telenor, Norway).

14. 3.30 - 5.30 ISSUE C (Research), session 1

Research Priorities in the Interface between Software and Performance (single session on Issue C to formulate recommendations to be discussed on Friday).

Summary and Final Discussion on Workshop Discussion Areas. Group reporter will summarize the draft recommendations for each issue; discussion will finalize them.

15. 8.30 - 9.30 ISSUE C (Research): Recommendations on Research Priorities in the Interface between Software and Performance.

16. 9.30 - 10.30 ISSUE B (Practice): Recommendations on Improving the State of Practice of Software Performance Engineering.

17. 11.00 - 12.00 ISSUE A: (Integration): Recommendations on how to Achieve Methodology Integration between Development and Performance Engineering

12.00 Close

8:30-10:15, Track 1, Performance Engineering of SDL/MSC Systems, A. Mitschele-Thiel, Universitat Erlangen, and Bruno Muller-Clostermann, Universitat-GH Essen.

The tutorial addresses issues related to the performance of systems specified with SDL and MSC and describes how performance aspects can be integrated with the SDL/MSC method. After an introduction into SDL and MSC, we identify the aspects that are relevant to performance evaluation and performance tuning. We introduce basic performance modeling techniques and survey approaches that integrate performance evaluation into the context of SDL and MSC.

10:45-12:30, Track 1, Software Performance Engineering and Simulation, R. Puigjaner and N. Vazquez, University of the Balearic Islands, Spain.

This tutorial aims at providing the fundamentals underlying a method for specifying hierarchical queuing network simulation models. Beginning with an overview of the state of the art of related areas (Software Performance Engineering, Object Oriented Method and Methodologies and Conceptual Simulation Models), the tutorial faces the problem of developing large scale simulation models in a multi team environment and presents HOOMA/SYS as a feasible method for leading with this type of scenarios. Finally, the tutorial presents the HELIOS research project as a real application of HOOMA/SYS in a multi team consortium in charge of developing a software to hardware hierarchy of performance models for second generation client/server systems.

1:30-3:15, Track 1, Design Patterns and Performance, D. Petriu, Carleton University, Canada.

Design patterns describe proven solutions to recurring design problems and promote quality characteristics in OO software such as flexibility, genericity, modularity, extensibility, maintainabiliy, etc. The use of patterns represents early design choices which may have a decisive influence on performance. Existing patterns cover various ranges of scale and abstraction, from high-level architectural patterns, to mid-level design patterns to low-level idioms. A pattern specifies two aspects of the solutio n it embodies: (i) the static structure, which is a configuration of cooperating classes (objects), and (ii) the run-time behaviour, which is expressed usually in the form of Message Sequence Charts (MSC).

Early performance analysis, such as recommended by the Software Performance Engineering methodology, has the potential to avoid performance disasters in complex OO designs. We can exploit the design patterns (or, more exactly, the abstractions they introduce at different levels) in order to facilitate this analysis.

The tutorial will discuss first the trade-offs between performance and other quality characteristics by the means of a few examples and measurements from the telecommunication domain. The second part will show how design patterns can facilitate the construction of Layered Queueing Networks performance models.

3:45-5:30, Track 1, Avoiding the Software Performance Crisis, M. Hesselgrave, Lucent Technologies

Members of the Bell Labs Software Technology Center Systems Architecture Group have used their experience as architecture consultants across Lucent Technologies (and its Bell System and AT&T predecessors) to identify common software project performance problems and to codify a methodology that addresses those problems. This process has been used successfully at Lucent on small projects with less than 10 people, and on large projects with development teams in many locations and countries. This 2-hour tutorial provides an overview of the methodology used to avoid performance crises at Lucent. It is intended for people interested in software development and its management. It helps students: understand the performance effort needed on projects; develop an awareness of software performance engineering; learn a process to predict, track, measure, and correct (if necessary) system performance.

8:30-12:30, Track 2, Unified Modeling Language (UML) and Software Performance Engineering, R. Pooley, University of Edinburgh.

For over a decade there have been proposals on how to integrate Performance Analysis into the Software Engineering process. These have included new languages and environments aimed at software designers and enhancements to software design techniques which include performance factors. Although some have been useful where used, none has been generally adopted so far.

Software designers seem to be agreeing on a common notation for expressing systems, the Unified Modelling Language (UML), backed by the Object management Group. As this becomes an industry standard, it seems timely to consider how to ensure that this is able to support performance engineering. A major revision of UML begins in October 1998 and ends in April 1999. The performance engineering community needs to sieze the opportunity to influence its outcome.

This tutorial presents:

• a short review of past work in performance engineering;
• an overview of UML and CASE tools which support it;
• how to integrate performance within the UML;
• how to influence the evolution of the UML.

The tutorial will be illustrated throughout with examples.

1:30-5:30, Track 2, Performance Engineering Evaluation of CORBA-based Distributed Systems, C. Smith, Performance Engineering Systems, and L. Williams, Software Engineering Research.

Systems using distributed object technology offer a number of advantages and their use is becoming more widespread. Distributed object systems are typically developed without regard to the locations of objects in the network or the nature of the communication between them. However, this approach often leads to performance problems due to latency in accessing remote objects and excessive overhead due to inefficient communication mechanisms. Thus, it is important to provide support for early assessment of the performance characteristics of such systems. Systems based on the OMG Object Management Architecture are relatively new and offer new challenges for performance modeling.

This tutorial first reviews SPE techniques for object-oriented systems focusing on early life-cycle issues; Use Cases serve as the bridge between object-oriented methods and SPE. Then it presents an overview of the CORBA object management performance issues. It reviews the OMA reference architecture and the Object Request Broker (ORB). Then it presents recent SPE extensions that address:

• performance engineering methods appropriate for distributed systems,
• extensions to the SPE modeling techniques to evaluate performance issues that arise when using an object request broker (ORB), and
• modeling techniques for CORBA synchronization primitives – both approximate models of software execution and advanced models of synchronization behavior.

A case study illustrates the techniques.

WOSP ’98

Tutorials and Workshop October 12-16, 1998

Please call or return this form directly to the hotel before midnight September 5, 1998. Reservations received after September 5 will be confirmed on a space-available basis at regular non-group rates. Ask for the room block under the name WOSP ’98. Please select a room type.

La Fonda Hotel, 100 E. San Francisco St.

Santa Fe, New Mexico 87501

Attention: Reservations

Reservations: 1-800-523-5002

Phone: 1-505-982-5511

Fax: 1-505-988-2952

Email: la-fonda@travelbase.com

Room rate: 138US, plus local taxes, currently 10.25%.

Single occupancy:_____ Double occupancy:_____

Smoking:_____ Non-smoking:_________

Arrival Date_________ Departure date___________

Arrival Time_________

Reservations are held until 4:00PM. To guarantee your room for arrival after 4:00 PM provide your credit card information below. A no-show charge is incurred unless you cancel more than 48 hours prior to the scheduled arrival date. Check in time is 3:00 PM and check out time is 12:00 PM.

Card type_______________Expiration Date___________

Card Number___________________________________

Signature______________________________________

Name_________________________________________

Company/Institute_______________________________

Address_______________________________________

City & State/Prov________________________________

Zip/Postal________________Country_______________

Daytime Phone___________Home Phone____________

If your require special arrangements or accommodations please let the hotel know of your specific needs in advance of your arrival.

Please complete both columns of this registration form. Payment must be made in US dollars. Please note the hotel registration cut-off date is September 5.

ACM Member [ ] $465 [ ] $545

Non-member [ ] $535 [ ] $615

Full time Student [ ] $150 [ ] $170

Choose track 1 or track 2_______________

ACM Member Number _____________

Special Dietary Meal Requested (please specify) ______________________________

Card type____________ Expiration Date___________

Card Number__________________________________

Signature_____________________________________

Name________________________________________

Title_________________________________________

Company/Institute______________________________

Address______________________________________

City & State/Prov______________________________

Zip/Postal____________Country__________________

Daytime Phone_____________________

Home Phone_______________________

Spouse's Name (if attending, optional) _____________________________________________

Payment may be made in US dollars by Credit Card Visa, Mastercard, or American Express, with a cheque drawn on an US bank, or with an international money order. Please send both sides of this form, and if applicable, a cheque or money order made payable to:

WOSP ’98

Registration

C/O Phillip Howard

11242 North 19^th Avenue

Phoenix, AZ, 85029, USA

Phone: 1-602-216-9100

Fax: 1-602-216-9200

Email: 74273.3427@compuserve.com

You may fax your registration if paying by credit card. There is no online registration.The conference registration fee includes attendance at all tutorial and technical sessions, a reception, the conference banquet, and one copy of the proceedings. Student registration does not include a ticket to the banquet.

In the event that you are unable to attend the meeting after you have paid fees, you will receive a full refund minus a $50 handling fee prior to October 9. No refunds will be available on or after October 9, but you will be sent a copy of the proceedings.

General Chair: Connie U. Smith (Performance Engineering Services)

Program Co-Chairs: Murray Woodside (Carleton University), Paul Clements (SEI)

Tutorials Chair: Pankaj Garg (HP Labs)

Registration: Phillip. C. Howard (Applied Computer Research)

Treasurer: Thad Jennings (Tivoli Systems)

Publicity: Jerry Rolia (Carleton University)

Exhibits: Janet Butler (Writer and Editor)

Heinz Beilner, Univ. of Dortmund

Giovanni Chiola, Univ. of Genoa

Gianfranco Ciardo, William and Mary

Asit Dan, IBM Watson Research Center

Larry Dowdy, Vanderbilt Univ.

Rich Friedrich, Hewlett-Packard Labs

Pankaj Garg, Hewlett-Packard Labs

Carlo Ghezzi, Politecnico di Milano

Neil Gunther, Performance Dynamics Consulting

Guenter Haring, Univ. of Vienna

Peter Harrison, Imperial College

Ravi Iyer, Univ. of Illinois

Anant Jalnapurkar, NORTEL

Mark Klein, SEI

Paul Kogut, Lockheed-Martin

Doug Lea, SUNY

John Lehoczky, CMU

Daniel Menasce, George Mason Univ.

Al Mok, Univ. of Texas

Gerard Mezaros, Solution Frameworks

Richard Muntz, UCLA

Gail Murphy, UBC

Rob Nord, Siemens

Harry Perros, North Carolina State

Dorina Petriu, Carleton University

Bill Riddle, SEI

Bill Sanders, Univ. of Illinois

Ken Sevcik, Univ. of Toronto

Doug Schmidt, Univ. of Washington

Satish Tripathi, UC Riverside

Vidar Vetland, Norwegian Telecom

Lloyd Williams, Software Eng. Research

Dave Zubrow, SEI