topic Re: low performance when running 'grep' and 'awk' scripting over large text file in Operating System - Linux

low performance when running 'grep' and 'awk' scripting over large text file

Alberto Rocha — Wed, 31 Jan 2007 16:54:21 GMT

Our application algorithm has to parse 12 times over a large text file (size=148MB).
We ran the same script over the same file on different machines and strangelly the worst performante was produced on our brand new server: an HP Proliant DL380G5 w/ 10krpm SAS disks. I suppose that it must be a question of tunning the kernel. What can i do to enhance the application processing time on this server?

TEST RESULTS (same script, same large file, different servers):
i) Laptop PC w/ Celeron processor runnig GentoOS (BEST PERFORM) - 1'38"
ii) HP Proliant DL380G3 Xeon 3GHz running RHEL3.6 - 6'14"
iii) HP Proliant DL380G5 2xXeon 2-core 3,4GHz running RHEL4.4 - 11'26"

Re: low performance when running 'grep' and 'awk' scripting over large text file

Ermin Borovac — Thu, 01 Feb 2007 03:02:28 GMT

This is a long shot by what is environment variable LANG set to?

In the past it's been reported that grep, sort and friends can be slow if LANG is set to one of the UTF variants.

Try

$ LANG=C grep ...

Re: low performance when running 'grep' and 'awk' scripting over large text file

Steven E. Protter — Thu, 01 Feb 2007 06:30:58 GMT

Shalom,

Interesting results.

The Os is different.

The processor is different.

The amount ofmemory is likely different.

The Patch level is different.

The storage(disk) performance is likely different.

Any of the factors above could impact the results.

More memory and a larger buffer cache will normally improve performance in such sitations. There are a lot of variables here.

SEP

Re: low performance when running 'grep' and 'awk' scripting over large text file

Alberto Rocha — Thu, 01 Feb 2007 07:41:30 GMT

BINGO! That is impressive: setting var env LANG, the execution time has reduced from 11' to 48". Thank you Ermin Borovac.