Hi, Does anyone use Seagate ST32000542AS disks with ZFS? I wonder if the performance is not that ugly as with WD Green WD20EARS disks. Thanks, -- Piotr Jasiukajtis | estibi | SCA OS0072 http://estseg.blogspot.com
On 29 November 2010 20:39, GMAIL <piotr.jasiukajtis at gmail.com> wrote:> Does anyone use Seagate ST32000542AS disks with ZFS? > > I wonder if the performance is not that ugly as with WD Green WD20EARS > disks. >I''m using these drives for one of the vdevs in my pool. The pool was created with ashift=12 (zpool binary from http://digitaldj.net/2010/11/03/zfs-zpool-v28-openindiana-b147-4k-drives-and-you/), which limits the minimum block size to 4KB, the same as the physical block size on these drives. I haven''t noticed any performance issues. These obviously aren''t 7200rpm drives, so you can''t expect them to match those in random IOPS. I''m also using a set of Samsung HD204UI''s in the pool. I would urge you to consider a 2^n + p number of disks. For raidz, p = 1, so an acceptable number of total drives is 3, 5 or 9. raidz2 has two parity drives, hence 4, 6 or 10. These vdev widths ensure that the data blocks are divided into nicer sizes. A 128KB block in a 9-wide raidz vdev will be split into 128/(9-1) = 16KB chunks. Cheers, -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://mail.opensolaris.org/pipermail/zfs-discuss/attachments/20101129/5aaf320d/attachment.html>
Thanks, I need to try modified zpool than. On Nov 29, 2010, at 10:50 AM, taemun wrote:> On 29 November 2010 20:39, GMAIL <piotr.jasiukajtis at gmail.com> wrote: > Does anyone use Seagate ST32000542AS disks with ZFS? > > I wonder if the performance is not that ugly as with WD Green WD20EARS disks. > > I''m using these drives for one of the vdevs in my pool. The pool was created with ashift=12 (zpool binary from http://digitaldj.net/2010/11/03/zfs-zpool-v28-openindiana-b147-4k-drives-and-you/), which limits the minimum block size to 4KB, the same as the physical block size on these drives. I haven''t noticed any performance issues. These obviously aren''t 7200rpm drives, so you can''t expect them to match those in random IOPS. > > I''m also using a set of Samsung HD204UI''s in the pool. > > I would urge you to consider a 2^n + p number of disks. For raidz, p = 1, so an acceptable number of total drives is 3, 5 or 9. raidz2 has two parity drives, hence 4, 6 or 10. These vdev widths ensure that the data blocks are divided into nicer sizes. A 128KB block in a 9-wide raidz vdev will be split into 128/(9-1) = 16KB chunks. > > Cheers, > _______________________________________________ > zfs-discuss mailing list > zfs-discuss at opensolaris.org > http://mail.opensolaris.org/mailman/listinfo/zfs-discuss-- Piotr Jasiukajtis | estibi | SCA OS0072 http://estseg.blogspot.com
On Mon, November 29, 2010 04:50, taemun wrote:> I would urge you to consider a 2^n + p number of disks. For raidz, p = 1, > so an acceptable number of total drives is 3, 5 or 9. raidz2 has two > parity drives, hence 4, 6 or 10. These vdev widths ensure that the data > blocks are divided into nicer sizes. A 128KB block in a 9-wide raidz vdev > will be split into 128/(9-1) = 16KB chunks.Wouldn''t nine disks in a a one-parity RAID set be pushing reliability a bit? Notwithstanding things like rebuild/resilver time and IOps, anyone know of a maximum recommended size to minimize the chances of losing an entire pool?
> I''m using these drives for one of the vdevs in my pool. The pool was created > with ashift=12 (zpool binary > from?http://digitaldj.net/2010/11/03/zfs-zpool-v28-openindiana-b147-4k-drives-and-you/), > which limits the minimum block size to 4KB, the same as the physical block > size on these drives. I haven''t noticed any performance issues. These > obviously aren''t 7200rpm drives, so you can''t expect them to match those in > random IOPS.The Seagate datasheet for those parts report 512-byte sectors. What is the deal with the ST32000542AS: native 512-byte sectors, native 4k-byte sector with selectable emulation, or native 4k-byte sectors with 512-byte sector emulation always on? Also, just a side note, I believe these drives achieve their "low-power" status with the reduced RPM (5900rpm), not with the head parking style power-management that WD Green drives use? The latter I''ve read is rather unsuitable for RAID operation (especially with HW RAID controllers).
On Mon, Nov 29, 2010 at 10:59 AM, Krunal Desai <movszx at gmail.com> wrote:> The Seagate datasheet for those parts report 512-byte sectors. What is > the deal with the ST32000542AS: native 512-byte sectors, native > 4k-byte sector with selectable emulation, or native 4k-byte sectors > with 512-byte sector emulation always on?Disregard; if I understand correctly, Seagate has proprietary "SmartAlign" tech that takes care of 4K sectors (see links below). I can''t seem to find any real whitepaper style explanation of the method though, but I assume it either: 1. does a really good job of 512-byte emulation that results in little to no performance degradation (http://consumer.media.seagate.com/2010/06/the-digital-den/advanced-format-drives-with-smartalign/ references "test data") 2. "dynamically" looks to see if it even needs to do anything; if the host OS is sending it requests that all 4k-aware/aligned, all is well. Newegg has these on sale today for $69.99; sadly the limit is 2. I think I''ll pick two up and use them for some tests and stock up on this model drive. Though, the power-on hours count seems rather "low" for me...8760 hours, or just 1 year of 24/7 operation. I may have to revisit power management in OpenSolaris (or upgrade to OpenIndiana) to see if my disks are spinning down when they are supposed too. Links: http://www.seagate.com/ww/v/index.jsp?locale=en-US&name=advanced-format-migration-to-4k-tpc&vgnextoid=746f43fce2489210VgnVCM1000001a48090aRCRD http://www.seagate.com/docs/pdf/whitepaper/tp615_smartalign_for_af_4k.pdf
On 30 November 2010 03:09, Krunal Desai <movszx at gmail.com> wrote:> I assume it either: > > 1. does a really good job of 512-byte emulation that results in little > to no performance degradation > ( > http://consumer.media.seagate.com/2010/06/the-digital-den/advanced-format-drives-with-smartalign/ > references "test data")2. "dynamically" looks to see if it even needs to do anything; if the> host OS is sending it requests that all 4k-aware/aligned, all is well. >My understanding is that this is merely saying that it will *align* the data correctly, with Windows XP, regardless of where Windows XP asks for the first sector to be. This has nothing to do with 512B random writes.> Though, the power-on hours count seems rather "low" > for me...8760 hours, or just 1 year of 24/7 operation.Not sure where you got this figure from, the "Barracuda Green" ( http://www.seagate.com/docs/pdf/datasheet/disc/ds1720_barracuda_green.pdf) is a different drive to the one we''ve been talking about in this thread ( http://www.seagate.com/docs/pdf/datasheet/disc/ds_barracuda_lp.pdf). I would note that the Seagate 2TB LP has a 0.32% Annualised Failure Rate. ie, in a given sample (which aren''t overheating, etc) 32 from every 10,000 should fail. I *believe* that the Power On-Hours on the Barra Green is simply saying that it is designed for 24/7 usage. It''s a per year number. I couldn''t imagine them specifying the number of hours before failure like that, just below an AFR of 0.43. Cheers, -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://mail.opensolaris.org/pipermail/zfs-discuss/attachments/20101130/240c841d/attachment.html>
> Not sure where you got this figure from, the "Barracuda Green" > (http://www.seagate.com/docs/pdf/datasheet/disc/ds1720_barracuda_green.pdf)?is > a different drive to the one we''ve been talking about in this thread > (http://www.seagate.com/docs/pdf/datasheet/disc/ds_barracuda_lp.pdf). > I would note that the Seagate 2TB LP has a 0.32% Annualised Failure Rate. > ie, in a given sample (which aren''t overheating, etc) 32 from every 10,000 > should fail. I *believe* that the Power On-Hours on the Barra Green is > simply saying that it is designed for 24/7 usage. It''s a per year number. I > couldn''t imagine them specifying the number of hours before failure like > that, just below an AFR of 0.43.Whoops, yes, that''s what I did, I assumed that LP == Green, but I guess that is not the case. I got 2 from the newegg sale, I''ll post my impressions once I get them and added to a pool...assuming they survived newegg''s rather subpar hard drive packaging process.
>>>>> "t" == taemun <taemun at gmail.com> writes:t> I would note that the Seagate 2TB LP has a 0.32% Annualised t> Failure Rate. bullshit. -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 304 bytes Desc: not available URL: <http://mail.opensolaris.org/pipermail/zfs-discuss/attachments/20101202/3c4c32f6/attachment.bin>
On 2 December 2010 16:17, Miles Nordin <carton at ivy.net> wrote:> >>>>> "t" == taemun <taemun at gmail.com> writes: > > t> I would note that the Seagate 2TB LP has a 0.32% Annualised > t> Failure Rate. > > bullshit. >Apologies, should have read: Specified Annualised Failure Rate. -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://mail.opensolaris.org/pipermail/zfs-discuss/attachments/20101202/a740e01d/attachment.html>