Grep versus FlatSQL versus MySQL
Queries using UNIX tools vs. a DBMS

by Konrad Lorincz, Kevin Redwine, Jesse Tov
Harvard University
January 2003

 

Abstract  |  Papers and Presentations  |  FlatSQL Implementations  |  Experiments

 

Abstract

We designed and implemented FlatSQL, a SQL-like query language for operating on flat files, and compared its performance to MySQL and grep. Our performance measurements help to determine which tool is appropriate under what circumstances. FlatSQL is convenient for many tasks because it works on text files and has SQL syntax, but its performance is inferior to MySQL for non-trivial indexed relations. However, we found some special cases where FlatSQL beats MySQL---very simple selects on un-indexed relations, and joins between un- indexed relations. For queries simple enough to be performed using grep, grep is faster than FlatSQL but cannot beat MySQL on indexed relations.

 

Papers and Presentations
  • Extended Abstract [pdf]
  • Final Paper [pdf]
  • Presentation [pdf] [ppt]

 

FlatSQL Implementations

 

Experiments