1 / 17

Oracle Measurement Results

Oracle Measurement Results. Prof. Bayer, PhD Dipl.-Inform. Volker Markl Roland Pieringer. Contents. Environment of the benchmark suite Results of the measurements Example measurement suite Comparison of the operating systems SUN Solaris and Windows NT Variation of the sizes of the tuples

morag
Download Presentation

Oracle Measurement Results

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Oracle Measurement Results Prof. Bayer, PhD Dipl.-Inform. Volker Markl Roland Pieringer

  2. Contents • Environment of the benchmark suite • Results of the measurements • Example measurement suite • Comparison of the operating systems SUN Solaris and Windows NT • Variation of the sizes of the tuples • Variation of the number of tuples • Variation of the number of restricted dimensions • Future Work

  3. Environment of the benchmark suite • Measures on two machines: • Compaq PC (4 Intel Processors, 200 MHz, 512 MB RAM) with Oracle 8.0.4.0.0 • SUN Ultra 2 (2 Ultra Sparc Processors, 200 MHz, 1 GB RAM) with Oracle 8.0.4.0.0 • Data uniformly distributed and created by a program • Elimination of cache side effects by NOCACHE

  4. Used indexes • UB: UB Tree (index that causes this meeting) • COMPOUND: concatenation of several indexes to get a multi dimensional indexin Oracle as index only table implemented • MULT: secondary indexes on all index attributes • SCAN:no index, relation scan, in Oracle as full table scan

  5. Q40% Q60% Q80% Q20% 20% 40% 60% 80% Kinds of benchmarks • c% measurements:Q = (I1,...,In)Ik = 1 % ..100 %I1, ..., Ik-1, Ik+1, In = c %random starting point • Cube measurements:Q=(I1,...,In)Ii = 0% .. 100%,for i = 1,...,nfix starting point

  6. Results of the measurements • Complete benchmarks suite • Comparison of Windows NT and SUN Solaris • Variation of the size of tuples • Variation of the number of tuples • Variation of the number of restricted dimensions

  7. 20% measure with 5 restricted dimensions(250K tuples, tuple size 428 Byte)

  8. 35%measure with 5 restricted dimensions(250K tuples, tuple size 428 Byte)

  9. Cube measure(250K tuples, tuple size 428 Byte)

  10. Solaris - NT:35% measure (125K tuples, 428 Byte tuple size) with SUN

  11. Solaris - NT:35% measure (125K tuples, 428 Byte tuple size) with NT

  12. Comparison of tuple size • Intersection of compound and relation scan: • small tuples: about 10% • large tuples: about 8% • Intersection UB Tree and relation scan: • small tuples: about 65% • large tuples: about 47% • Decrease of performance of the relation scan: • small tuples: factor 4 • large tuples: factor 1.4

  13. Comparison of the number of tuplesCube measure (1M tuples, 228 Byte tuple size)

  14. Comparison of the number of tuplesCube measure (2M tuples, 228 Byte tuple size)

  15. Comparison of the number of tuplesCube measure (4M tuples, 228 Byte tuple size)

  16. Variation of the number of restricted dimensionsCube measure (250K tuples, 428 Byte tuple size)

  17. Variation of the number of restricted dimensions35% measure (250K tuples, 428 Byte tuple size)

More Related