S1: Csci 8701 - Week 2
- Administrative
- HW 1 - any questions
- Project selection - any questions
- Sources - DBLP (papers), amazon.com (books)
- Goals:
- Understand benchmarking
- Review TP1, Sequoia 2000
- Recommended Readings:
- M. Stonebraker et al.
- - The Sequoia 2000 Storage Benchmark
- Anon et al.
- -A Measure of Transaction Processing Power
- Project Ideas:
- Benchmarks for e-commerce, CRM, Data Mining
Copyright: S. Shekhar, C. S. Dept.,
University of Minnesota, Minneapolis, MN 55455. ,,.(home)
S2: TP1 (Anon et al.)
- Problem Statement
- Compare OLTP systems on price, performance
- Evaluate I/O architecture (not just CPU)
- Major Contributions
- Consensus workload and measures
- DebitCredit benchmark (data, xaction)
- TPS measures of performance (see pp. 613)
- Key Concepts in OLTP
- 3 transaction types - interactive, minibatch, bulk
- What to test?: hardware, library, logs, locking (hotspots)
- Q? Match system part exercised by each transaction type:
- I/O hardware, I/O run-time library, locking, log
Copyright: S. Shekhar, C. S. Dept.,
University of Minnesota, Minneapolis, MN 55455. ,,.(home)
S3: TP1 (Anon et al.)
- Validation Methodology
- Prototyping, experiments, measurements
- Consensus building, discussion, self-critique
- Assumptions
- Performance: startup, communication,
- Value of clustering, index selection, distributed DB
- Dollar Costs: outage, deployment
- Rewrite today
- Use standard SQL syntax
- Main memory caching requirements
- Use e-commerce application
- See TPC-C, TPC-D
Copyright: S. Shekhar, C. S. Dept.,
University of Minnesota, Minneapolis, MN 55455. ,,.(home)
S4: Sequoia 2000 (Stonebraker)
- Problem Statement
- Compare DBMS for scientific workload
- Evaluate object relational features
- Major Contributions
- Made a scientific dataset public
- Object relational data model, queries
- Key Concepts in Earth Science Datasets
- What to test?: spatial index/ADTs/functions
- 5 data types - rastor, points, lines, polygons, DAGs
- 4 query types - bulk load, selection (range query),
- ... spatial join, and transitive closure
- Q? Match eleven queries in paper to query types above.
Copyright: S. Shekhar, C. S. Dept.,
University of Minnesota, Minneapolis, MN 55455. ,,.(home)
S5: Sequoia 2000 (Stonebraker)
- Validation Methodology
- Prototyping, experiments, measurements
- Collaboration with Domain scientists
- Assumptions
- Client-server division of work: Spatial ADTs
- Value of update, locks, logs,
- Dollar Costs: outage, deployment
- Rewrite today
- Use SQL1999 (SQL3) syntax, OGIS ADTs
- Candidate sets - ESRI SDE, Oracle SDE,
- Expand components - e.g. update
- See Spatial Database Book
Copyright: S. Shekhar, C. S. Dept.,
University of Minnesota, Minneapolis, MN 55455. ,,.(home)
S6: Ch. 1: Introduction to Spatial Databases
- Spatial databases
- Lot more than Sequoia 2000 (Stonebraker)
- 1.2 Applications beyond Earth Science
- 1.6 A stroll through Spatial Databases
- Space Taxonomy, Data Models
- Query Language: OGIS datatypes (in SQL3 or Java)
- Query Processing: filter-and-refine
- File organization, Indexes
- Query Processing: cost models, strategies
- Graph datatypes and queries (mapquest)
- Rastor queries (map algebra)
Copyright: S. Shekhar, C. S. Dept.,
University of Minnesota, Minneapolis, MN 55455. ,,.(home)