William Alexander
ABSTRACT
The same application implemented on different systems will necessarily present different workloads to the systems. Characterizations of workloads intended to represent the same application, but input to models of different systems, must also differ in analogous ways. We present a hierarchical method for characterizing a workload at increasing levels of detail such that every characterization at a lower level still accurately represents the workload at higher levels.
We discuss our experience in using the method to feed the same application through a workload characterization “pipeline” to two different models of two different systems, a conventional relational database system and a logic-based distributed database system. We have developed programs that partially automate the characterization changes that are required when the system to be modeled changes.
- Anon85.Anon. et al, "A Measure of Transaction Processing Power", Datamation, April 1, 1985, pp 112-118. Google Scholar
Digital Library
- Ban87.Francois Bancilhon, Ted Briggs, Setrag Khoshafian, and Patrick Valduriez, "FAD, a Powerful and Simple Database Language", Submitted to 13th VLDB Conference, Brighton, UK, September 1987. Google ScholarDigital Library
- Bri82.R. S. Brice, "Benchmarking Your Benchmarks", Proc. of the Computer Measurement Group (CMG) XIII International Conf., December 1982, San Diego, CA.Google Scholar
- Car75.A.F. Cardenas, "Analysis and Performance of Inverted Database Structures", CACM, Vol. 18, No. 5 (May i975), pp 253-263. Google ScholarDigital Library
- Che84.j.M. Cheng, C. R. Loosley, A. Shibamiya, and P. S. Worthington, "IBM Database 2 Performance: Design, Implementation, and Tuning", IBM Systems Journal, Vol. 23, No. 2, 1984, pp 189-210.Google ScholarDigital Library
- Den78.Peter J. Denning and Jeffrey P. Buzen, "The Operational Analysis of Queueing Network Models", Computing Surveys 10, 3 (September 1978), pp 225-261. Google ScholarDigital Library
- Laz84.Edward D. Lazowska, John Zahorjan, G. Scott Graham, and Kenneth C. Sevcik, Quantitative System Performance, Prentice-Hall, 1984. Google ScholarDigital Library
- Sev81.Kenneth C. Sevcik, "Data Base System Performance Prediction Using An Analytical Model", Proceedings of 7th VLDB Conf., 1981, pp 182-189.Google Scholar
- Sch85.M. Schkolnick and P. Tiberio, "Estimating the Cost of Updates in Relational Database", A CM TODS, Vol. 10, No. 2 (June 1985), pp 163-179. Google ScholarDigital Library
- Schw85.H. Schwetman, "CSIM: A C-Based, Process-Oriented Simulation Language", MCC Parallel Processing Technical Report No. PP-080-85, September 1985.Google Scholar
- Smi80.Connie U. Smith, "The Prediction and Evaluation of the Performance of Software from Extended Design Specifications", PhD Dissertation, University of Texas at Austin, University Microfilms Pub. No. KRA81-00963, August 1980. Google ScholarDigital Library
- Smi82.Connie U. Smith and J. C. Browne, "Performance Engineering of Software Systems: A Case Study", Proc. National Computer Conference, Vol 15, Houston, June 1982, pp 217-224.Google ScholarDigital Library
- Tsu86.Shalom Tsur and Carlo Zaniolo, "LDL: A Logic-Based Data-Language", Proc. of 12th VLDB Conference, August 1986, Kyoto, pp 33-41. Google ScholarDigital Library
Index Terms
- A workload characterization pipeline for models of parallel systems
Recommendations
A workload characterization pipeline for models of parallel systems
The same application implemented on different systems will necessarily present different workloads to the systems. Characterizations of workloads intended to represent the same application, but input to models of different systems, must also differ in ...
Workload characterization: A key aspect of microarchitecture design
Micro editor in chief Pradip Bose talks about the issue of workload characterization in the modern setting.
Workload Characterization: Survey of Current Approaches and Research Challenges
ICCCT-2017: Proceedings of the 7th International Conference on Computer and Communication TechnologyWorkload is a set of inputs given to a infrastructure for processing. Performance can be measured based on the efficient processing of the workloads. Different workloads has different set of characteristics. In this paper, we have mainly focused on ...
Comments