From: Haessner, Roland (Roland.Haessner@t-systems.com)
Date: Sun Jul 14 2002 - 13:54:18 EDT
Hi admins :-)
I have a question to you :
We are using an oracle8 database, HPUX 11i, a L-Cass server (2 CPU, 4GB RAM)
and a VA7100 as storage system, connected via two brocade switches. When we
are using a dd-command from /dev/zero to the LUN's (50 MB file), we get
different time values :
- with a blocksize of 1024 (bs=1024) it needs about 2 seconds ... that's
okay
- with a blocksize of 1000 (bs=1000) it needs about 40 seconds !
That's crazy, isn't it ? A similar behaviour shows the tar-command.
To analyse the performance and for understanding the optimal parameter
values for buffer cache, kernel parameter and other io specific parameters i
would like to know more about the way of IO ... from the application down to
the disks !
Are there any whitepapers, overviews or descriptions about this way of IO,
about main parameters ?
Do you know where this big difference comes from ? Why is this blocksize
soooo impotant ? How can i check the values of these parameters ?
Thanks for all answers to my mail, i will summarize as soon as possible !
Roland
:-)
-- ---> Please post QUESTIONS and SUMMARIES only!! <--- To subscribe/unsubscribe to this list, contact majordomo@dutchworks.nl Name: hpux-admin@dutchworks.nl Owner: owner-hpux-admin@dutchworks.nl Archives: ftp.dutchworks.nl:/pub/digests/hpux-admin (FTP, browse only) http://www.dutchworks.nl/htbin/hpsysadmin (Web, browse & search)
This archive was generated by hypermail 2.1.7 : Sat Apr 12 2008 - 11:02:16 EDT