i/o benchmark: vxbench

[Klorese, Roger B.A.]

It's on our web site -- see
http://www.veritas.com/full/bench/fs/vxbench.tar

>-----Original Message-----
>From:	Fred True [SMTP:ft@research.att.com]
>Sent:	Thursday, March 19, 1998 11:17 AM
>To:	SSA Managers
>Subject:	i/o benchmark: vxbench
>
>
>To whoever was looking for one of these (sorry, I misplaced the original
>msg); if you have a relationship with veritas or are using vxvm, ask them
>for a copy of vxbench.  It's a simple tool but can prove quite useful at
>demonstrating device or filesystem performance under various i/o loads.
>
>It allows sequential or random i/o; read, write or mixed tests, and has a
>bunch of knobs:
>
>        "nrep=n            repeat the I/O loop in the test n times",
>        "nthreads=n        number of threads accessing each file",
>        "iosize=n          size of each I/O",
>        "fsync             do an fsync on the file after writing it",
>        "remove            remove each file after the test",
>        "iocount=n         number of I/Os",
>        "reserveonly       reserve space for the file but don't do I/O",
>        "maxfilesize       maximum offset in K for random I/O tests",
>        "randseed          seed value for random number generator",
>        "truncup           set an inital file size for random I/O",
>        "rdpct=n           set read percentage of job mix for mixed tests",
>
>Perhaps someone from veritas could post this to the list; or give me
>permission to do so?
>
>Fred
>
>--
>Fred True				"My name is Ozymandias, King of Kings:
>AT&T Labs Network Mathematics Research	 Look on my works, ye Mighty,
>ft@research.att.com			 and despair!"
>						-P. B. Shelley
>