From: <Mar...@De...> - 2008-08-21 16:15:17
|
I am using a raw disk. I am using 63MB requests, because if I use 64MB requests, then IOMETER errors. I assumed that 63MB was the biggest transfer that could be handled by a SATA drive. There are limits to the amount of data that can be transferred, in one command, at the drive interface. I tried a 100GB transfer size but it would not work. The "steps" I mentioned are what I should have referred to as a test. I thought I might be able to string together a series of 63MB tests and cover the entire disk that way. So the script has a series of 63MB Sequential Write tests. As IOMETER steps through each 63MB "test" will it proceed from where it left off or will the next test restart ant LBA 0? Mark Lindholm Enterprise HDD Engineering 512 725 3195 mar...@de... From: ming zhang [mailto:min...@gm...] Sent: Thursday, August 21, 2008 7:20 AM To: Vedran Degoricija Cc: Lindholm, Mark; iom...@li... Subject: Re: [Iometer-user] Filling a drive raw disk is good. over file system is bad, (not real sequential at disk level) On Thu, Aug 21, 2008 at 12:26 AM, Vedran Degoricija <ve...@ya...> wrote: Mark, Why are you using ~64MB requests? Whatever request size you select, as long as you have chosen a sequential I/O pattern w/out an upper bound LBA limit, *I think* it will start at LBA 0 and go to the end. Whatever the right answer is, it should apply to a raw disk or a file system. Ved ----- Original Message ---- From: "Mar...@De..." <Mar...@De...> To: iom...@li... Sent: Wednesday, August 20, 2008 3:11:58 PM Subject: Re: [Iometer-user] Filling a drive Actually the pseudo-random that IOMETER does is good enough for me. The question now is If I write the script to do repeated steps of 63MB 100%Sequential 100%Write. What will happen? I assume IOMETER will start at LBA 0 and write the 64MB sequentially as long as I allow it enough time. What will happen on the next step? Will IOMETER go back to LBA 0 or will it continue from where it left off (LBA 123046) in the prior step? I know it will be Sequential within each step of 63MB, but will it go back to LBA 0 when I go to the next step to tranfer the next 63MB of data? Mark Lindholm Enterprise HDD Engineering 512 725 3195 mar...@de... From: iom...@li... [mailto:iom...@li...] On Behalf Of ming zhang Sent: Wednesday, August 20, 2008 3:21 PM To: Jan...@As... Cc: IOmeter Mailing List; Lindholm, Mark Subject: Re: [Iometer-user] Filling a drive yes, that is the way to get large # of random data, except it is slow... On Wed, Aug 20, 2008 at 4:14 PM, Jan...@As... <Jan...@as...> wrote: Just use the random number device available in most Linux distros: The example below creates a 20KB random file. You can leave out the count argument to fill up the whole filesystem. Specifying of=/dev/sdb will fill up the 2nd disk attached to your system. You don't need a partition or filesystem for this. ~$ ls /dev/urandom /dev/urandom ~$ dd if=/dev/urandom of=/tmp/20K bs=1024 count=20 20+0 records in 20+0 records out 20480 bytes (20 kB) copied, 0.0190392 s, 1.1 MB/s ~$ ls -l /tmp/20K -rw-r--r-- 1 jan jan 20480 2008-08-20 21:52 /tmp/20K ~$ Rgds, Jan. On Wed, 2008-08-20 at 15:57 -0400, ming zhang wrote: it only fill a buffer with random and then repeat the same data again and again... or maybe that is good enough for you already.... looks like a lot of people want a raw disk level benchmark tool.... ps, always cc to iometer list... On Wed, Aug 20, 2008 at 3:51 PM, <Mar...@de...> wrote: I believe IOMETER uses pseudo-random and that is good enough for my purpose. If not what does IOMETER write? Mark Lindholm Enterprise HDD Engineering 512 725 3195 mar...@de... From: ming zhang [mailto:min...@gm...] Sent: Wednesday, August 20, 2008 2:47 PM To: Lindholm, Mark Cc: iom...@li... Subject: Re: [Iometer-user] Filling a drive On Wed, Aug 20, 2008 at 2:37 PM, <Mar...@de...> wrote: I want to use an IOMETER script to fill a SATA or SAS drive with random data. i do not think iometer will write true random data. so this might not meet your purpose at all... The biggest transfer that I can make at one time appears to be something less than 64MB. If I write the script to do repeated steps of 63MB 100%Sequential 100%Write. What will happen? I assume IOMETER will start at LBA 0 and write the 64MB sequentially as long as I allow it enough time. What will happen on the next step? Will IOMETER go back to LBA 0 or will it continue from where it left off (LBA 123046) in the prior step? left off. otherwise it is not sequential at all. Mark Lindholm Enterprise HDD Engineering 512 725 3195 mar...@de... ------------------------------------------------------------------------ - This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/ _______________________________________________ Iometer-user mailing list Iom...@li... https://lists.sourceforge.net/lists/listinfo/iometer-user ------------------------------------------------------------------------ - This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/ _______________________________________________ Iometer-user mailing list Iom...@li... https://lists.sourceforge.net/lists/listinfo/iometer-user ------------------------------------------------------------------------ - This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/ _______________________________________________ Iometer-user mailing list Iom...@li... https://lists.sourceforge.net/lists/listinfo/iometer-user |