Re: [ADSM-L] DISK I/O Blocksize

2007-04-18 08:59:19
Subject: Re: [ADSM-L] DISK I/O Blocksize
From: Ibán Bernaldo de Quirós Márquez <Iban.Bernaldodequiros AT SUN DOT COM>
Date: Wed, 18 Apr 2007 14:58:41 +0200
Thinking of the technote provided by Richad Sims (Thanks!!) don't you think that the segment size for disk stgpool is very small ¿? 4KB ¿?
If I have a RAID 5 (4+1 volumes), configured with 256 KBsegment size, am
I not wasting space... ¿?
If one client is doing backup, the backup(objects) are coming across the
lan and received by the ethernet adapter of the TSM server. These
objects are going to main memory of the TSM server (processed), and
after they are flushed to disk in segments more big than 4 KB (perhaps
256 KB) ¿? or it goes directly to disk without passing first to memory ¿?
So does TSM have a multiplier like oracle db to read more blocks
sequentially ¿?
If I have large I/O requests it seems to me that 4KB for blocksize is
very small... or perhaps am I forgetting something about TSM
functionality... some tuning parameters ¿?
Does TSM do a caching or something previous to write to disk and tape ¿?

Thanks In Advance !!


Richard Sims wrote:
On Apr 16, 2007, at 6:07 AM, Ibán Bernaldo de Quirós Márquez wrote:

Hi all,

Did anyone know what I/O blocksize TSM writes to disk ¿? TSM database writes 4 KB but what about STGPOOLs, and RLOG ¿?
See IBM Technote 1240946.

<Prev in Thread] Current Thread [Next in Thread>