Database Benchmark: Unterschied zwischen den Versionen

Aus Geoinformation HSR
Wechseln zu: Navigation, Suche
K
K
Zeile 16: Zeile 16:
  
 
A database performance benchmark hast to consider following different aspects:
 
A database performance benchmark hast to consider following different aspects:
# Kalt- und Warmstart, und beim Warmstart ist der Unterschied zwischen der 1. Query und der 2. Query besonders interessant (Caching der DB selber).
+
# Cold and warm start (beware that in case of warm start chaching will take place!).
# Equality und Range Queries.
+
# Equality and Range Queries.
# Query-Resultsets, die ein Tupel liefen und solche die über die Hälfte der DB zurückliefern.
+
# Query-Result Sets, which respond with one tupel or which respond more than half of the tupels in the dataset.
 
# Single User versus Multi-user.
 
# Single User versus Multi-user.
  
Software (Scritps) zur Automatisierung von Benchmarks:
+
Software (Scritps) for benchmark automation:
 
* [http://wiki.hsr.ch/Datenbanken/files/db-benchmark_mott.zip PostgreSQL hstore Benchmark] - Benchmarking in Python by Michel Ott, 2011.
 
* [http://wiki.hsr.ch/Datenbanken/files/db-benchmark_mott.zip PostgreSQL hstore Benchmark] - Benchmarking in Python by Michel Ott, 2011.
 
* [http://www.postgresql.org/docs/devel/static/pgbench.html pgbench] - Benchmark tool for PostgreSQL.
 
* [http://www.postgresql.org/docs/devel/static/pgbench.html pgbench] - Benchmark tool for PostgreSQL.

Version vom 13. Oktober 2013, 20:48 Uhr

See also

About Database Performance Benchmarking...

Existing DB-Benchmarks:

  • TPC-C for OLTP benchmarks.
  • TPC-R & TPC-H (formerly TPC-DS) for data warehouse & decision support systems.
  • TPC-W benchmark for Web-based systems.
  • "The Engineering Database Benchmark".
  • Open Source Database Benchmark
  • PolePosition open source database benchmark [1]

Guidelines

A database performance benchmark hast to consider following different aspects:

  1. Cold and warm start (beware that in case of warm start chaching will take place!).
  2. Equality and Range Queries.
  3. Query-Result Sets, which respond with one tupel or which respond more than half of the tupels in the dataset.
  4. Single User versus Multi-user.

Software (Scritps) for benchmark automation:

Weblinks