Advanced

Performance analysis and improvement of PostgreSQL

Lindblom, Martin LU and Strandin, Fredrik LU (2015) EDA920 20142
Department of Computer Science
Abstract
PostgreSQL is a database management system, used in many different applications throughout the industry. As databases often are the bottlenecks in the performance of applications, their performance becomes crucial. Better performance can either be achieved by using more and faster hardware, or by making the software more efficient. In this master thesis we do a performance analysis of the PostgreSQL database server from the perspective of compiler optimizations, file systems and software prefetching. We will also show how a data structure used in PostgreSQL can benefit from manually introducing software prefetching, as it is hard for the compiler to predict cache misses and insert prefetching instructions in a profitable way.
Popular Abstract (Swedish)
PostgreSQL är en populär databasserver som används i stora delar av industrin. Databasservrar används för att lagra, bearbeta och hämta uppgifter åt t.ex. företag, organisationer och universitet som de förlitar sig på i sitt arbete. Ofta blir dock dessa databaser flaskhalsen i hur snabbt arbete kan utföras, och därför har vi analyserat och förbättrat en populär databasserver.
Please use this url to cite or link to this publication:
author
Lindblom, Martin LU and Strandin, Fredrik LU
supervisor
organization
course
EDA920 20142
year
type
H3 - Professional qualifications (4 Years - )
subject
keywords
PostgreSQL, postgres, performance analysis, optimization, prefetching, filesystems
report number
2015-01
ISSN
1650-2884
language
English
id
4939915
date added to LUP
2015-01-26 10:01:06
date last changed
2015-02-24 14:14:46
@misc{4939915,
  abstract     = {PostgreSQL is a database management system, used in many different applications throughout the industry. As databases often are the bottlenecks in the performance of applications, their performance becomes crucial. Better performance can either be achieved by using more and faster hardware, or by making the software more efficient. In this master thesis we do a performance analysis of the PostgreSQL database server from the perspective of compiler optimizations, file systems and software prefetching. We will also show how a data structure used in PostgreSQL can benefit from manually introducing software prefetching, as it is hard for the compiler to predict cache misses and insert prefetching instructions in a profitable way.},
  author       = {Lindblom, Martin and Strandin, Fredrik},
  issn         = {1650-2884},
  keyword      = {PostgreSQL,postgres,performance analysis,optimization,prefetching,filesystems},
  language     = {eng},
  note         = {Student Paper},
  title        = {Performance analysis and improvement of PostgreSQL},
  year         = {2015},
}