Hi there I know the above mentioned kit (2530) is new, but has anybody tried a direct attached SAS setup using zfs? (and the Sun SG-XPCIESAS-E-Z card, 3Gb PCI-E SAS 8-Port Host Adapter, RoHS:Y - which is the prefered HBA I suppose) Did it work correctly? Thank you
Louwtjie Burger wrote:> I know the above mentioned kit (2530) is new, but has anybody tried a > direct attached SAS setup using zfs? (and the Sun SG-XPCIESAS-E-Z > card, 3Gb PCI-E SAS 8-Port Host Adapter, RoHS:Y - which is the > prefered HBA I suppose) > Did it work correctly?Yes, it was tested as part of our project to add support to mpt for MPxIO. The zfs test suite was one of the required tests in our suite. Yes, it worked correctly. What other questions do you have? cheers, James C. McPherson -- Solaris kernel software engineer Sun Microsystems
On 5/30/07, James C. McPherson <James.McPherson at sun.com> wrote:> Louwtjie Burger wrote: > > I know the above mentioned kit (2530) is new, but has anybody tried a > > direct attached SAS setup using zfs? (and the Sun SG-XPCIESAS-E-Z > > card, 3Gb PCI-E SAS 8-Port Host Adapter, RoHS:Y - which is the > > prefered HBA I suppose) > > Did it work correctly? > > Yes, it was tested as part of our project to add support to > mpt for MPxIO. The zfs test suite was one of the required > tests in our suite. > > Yes, it worked correctly.Just to confirm... you connected a Galaxy AMD server to a 2530JBOD, using the SAS controller from Sun... you could see the SAS/SATA disks (without a proper 2530 controller unit)? Thanks James
Hello Louwtjie, Monday, June 4, 2007, 9:14:26 AM, you wrote: LB> On 5/30/07, James C. McPherson <James.McPherson at sun.com> wrote:>> Louwtjie Burger wrote: >> > I know the above mentioned kit (2530) is new, but has anybody tried a >> > direct attached SAS setup using zfs? (and the Sun SG-XPCIESAS-E-Z >> > card, 3Gb PCI-E SAS 8-Port Host Adapter, RoHS:Y - which is the >> > prefered HBA I suppose) >> > Did it work correctly? >> >> Yes, it was tested as part of our project to add support to >> mpt for MPxIO. The zfs test suite was one of the required >> tests in our suite. >> >> Yes, it worked correctly.LB> Just to confirm... you connected a Galaxy AMD server to a 2530JBOD, LB> using the SAS controller from Sun... you could see the SAS/SATA disks LB> (without a proper 2530 controller unit)? Can someone confirm it (that connecting 2530 expansion tray/JBOD directly to Solaris x86 using SAS cards + MPxIO works)? -- Best regards, Robert mailto:rmilkowski at task.gda.pl http://milek.blogspot.com
Robert Milkowski wrote:> Hello Louwtjie, > > Monday, June 4, 2007, 9:14:26 AM, you wrote: > > LB> On 5/30/07, James C. McPherson <James.McPherson at sun.com> wrote: >>> Louwtjie Burger wrote: >>>> I know the above mentioned kit (2530) is new, but has anybody tried a >>>> direct attached SAS setup using zfs? (and the Sun SG-XPCIESAS-E-Z >>>> card, 3Gb PCI-E SAS 8-Port Host Adapter, RoHS:Y - which is the >>>> prefered HBA I suppose) >>>> Did it work correctly? >>> Yes, it was tested as part of our project to add support to >>> mpt for MPxIO. The zfs test suite was one of the required >>> tests in our suite. >>> >>> Yes, it worked correctly. > > LB> Just to confirm... you connected a Galaxy AMD server to a 2530JBOD, > LB> using the SAS controller from Sun... you could see the SAS/SATA disks > LB> (without a proper 2530 controller unit)? > > Can someone confirm it (that connecting 2530 expansion tray/JBOD > directly to Solaris x86 using SAS cards + MPxIO works)?Hi all, sorry for the late response (buying a house turns out to be a stressful thing!) Yes, my team''s test plan did include ST2530 array attached to SAS hba. As far as I understand it, I do not think that a plain jbod version of the ST2530 is supported. I believe that a jbod attached to the ST2540 (fc-connected) is supported. Louwtjie - have you tried a ST2530 as a jbod and found it lacking? If so, please let me know the details directly. cheers, James C. McPherson -- Solaris kernel software engineer Sun Microsystems
Hello James, Wednesday, June 13, 2007, 1:06:22 PM, you wrote: JCM> Robert Milkowski wrote:>> Hello Louwtjie, >> >> Monday, June 4, 2007, 9:14:26 AM, you wrote: >> >> LB> On 5/30/07, James C. McPherson <James.McPherson at sun.com> wrote: >>>> Louwtjie Burger wrote: >>>>> I know the above mentioned kit (2530) is new, but has anybody tried a >>>>> direct attached SAS setup using zfs? (and the Sun SG-XPCIESAS-E-Z >>>>> card, 3Gb PCI-E SAS 8-Port Host Adapter, RoHS:Y - which is the >>>>> prefered HBA I suppose) >>>>> Did it work correctly? >>>> Yes, it was tested as part of our project to add support to >>>> mpt for MPxIO. The zfs test suite was one of the required >>>> tests in our suite. >>>> >>>> Yes, it worked correctly. >> >> LB> Just to confirm... you connected a Galaxy AMD server to a 2530JBOD, >> LB> using the SAS controller from Sun... you could see the SAS/SATA disks >> LB> (without a proper 2530 controller unit)? >> >> Can someone confirm it (that connecting 2530 expansion tray/JBOD >> directly to Solaris x86 using SAS cards + MPxIO works)?JCM> Hi all, JCM> sorry for the late response (buying a house turns out JCM> to be a stressful thing!) JCM> Yes, my team''s test plan did include ST2530 array attached JCM> to SAS hba. But there''s 2530 with RAID controller and SAS external ports. To clarify I was asking about expansion trays without any RAID controllers - just 2530 jbod attached with dual links to a host + MPxIO. Can you confirm you did such tests and it works? Sorry.. but I would like to confirm it before buying. JCM> As far as I understand it, I do not think that a plain JCM> jbod version of the ST2530 is supported. I believe that JCM> a jbod attached to the ST2540 (fc-connected) is supported. If it works it doesn''t have to be supported. But imho it should be. -- Best regards, Robert mailto:rmilkowski at task.gda.pl http://milek.blogspot.com
Robert Milkowski wrote: ...> JCM> Yes, my team''s test plan did include ST2530 array attached > JCM> to SAS hba. > > But there''s 2530 with RAID controller and SAS external ports. > To clarify I was asking about expansion trays without any RAID > controllers - just 2530 jbod attached with dual links to a host + > MPxIO. > Can you confirm you did such tests and it works? > Sorry.. but I would like to confirm it before buying.As I dug deeper into the testing reports, it appears that a ST2530 jbod config wasn''t tested. Certainly it is the case that I do not see an ST2530 jbod mentioned in the latest release notes document that I have reviewed.> JCM> As far as I understand it, I do not think that a plain > JCM> jbod version of the ST2530 is supported. I believe that > JCM> a jbod attached to the ST2540 (fc-connected) is supported. > > If it works it doesn''t have to be supported.If it''s sold by Sun then it has to be supported. It''s one of those contractual obligation things. As always, please confer with your Sun sales rep for the actual details. All I''ve done is work on the software for the product :-) James C. McPherson -- Solaris kernel software engineer Sun Microsystems
Robert Milkowski wrote: ...> JCM> As far as I understand it, I do not think that a plain > JCM> jbod version of the ST2530 is supported. I believe that > JCM> a jbod attached to the ST2540 (fc-connected) is supported. > > If it works it doesn''t have to be supported.and practically speaking, I expect that it would just work. James C. McPherson -- Solaris kernel software engineer Sun Microsystems
On Wed, 13 Jun 2007, James C. McPherson wrote:> Robert Milkowski wrote: > ... >> JCM> As far as I understand it, I do not think that a plain >> JCM> jbod version of the ST2530 is supported. I believe that >> JCM> a jbod attached to the ST2540 (fc-connected) is supported. >> >> If it works it doesn''t have to be supported. > > and practically speaking, I expect that it would > just work.Is there any chance you could arm-twist someone into running a quick test? Regards, Al Hopper Logical Approach Inc, Plano, TX. al at logical-approach.com Voice: 972.379.2133 Fax: 972.379.2134 Timezone: US CDT OpenSolaris Governing Board (OGB) Member - Apr 2005 to Mar 2007 http://www.opensolaris.org/os/community/ogb/ogb_2005-2007/
Al Hopper wrote:> On Wed, 13 Jun 2007, James C. McPherson wrote: > >> Robert Milkowski wrote: >> ... >>> JCM> As far as I understand it, I do not think that a plain >>> JCM> jbod version of the ST2530 is supported. I believe that >>> JCM> a jbod attached to the ST2540 (fc-connected) is supported. >>> >>> If it works it doesn''t have to be supported. >> >> and practically speaking, I expect that it would >> just work. > > Is there any chance you could arm-twist someone into running a quick test?I''ll see what I can do. No promises though. cheers, James C. McPherson -- Solaris kernel software engineer Sun Microsystems
On Thu, 14 Jun 2007, James C. McPherson wrote:> Al Hopper wrote: >> On Wed, 13 Jun 2007, James C. McPherson wrote: >> >>> Robert Milkowski wrote: >>> ... >>>> JCM> As far as I understand it, I do not think that a plain >>>> JCM> jbod version of the ST2530 is supported. I believe that >>>> JCM> a jbod attached to the ST2540 (fc-connected) is supported. >>>> >>>> If it works it doesn''t have to be supported. >>> >>> and practically speaking, I expect that it would >>> just work. >> >> Is there any chance you could arm-twist someone into running a quick test? > > I''ll see what I can do. No promises though.[ .... offlist .... ] Thanks. Many Thanks. It seems wasteful to (determine the required part number and) order a 2530 JBOD "expansion" shelf and then return it if it does not work out.... Obviously this is a huge hole in Suns'' current storage product line. There is a huge need for a ZFS saavy JBOD "box" with a high speed host interconnect. The nearest alternative I''ve been able to dig up is the Promise VTrak J300S - but I''d rather buy a Sun (or StorageTek) part than a Promise Tech part. And, regardless of what the profit margin is, I''d rather see Sun with that profit margin than Promise Tech. Why? .... because: a) I want to see Sun/OpenSolaris persist into the far distant future. b) I''m a Sun customer - not a Promise Tech customer. c) Sun engineers products - Promise is a mass market, lowest-common-demoninator type supplier - with loyalty only to those (upstream) suppliers who can provide the highest market volume opportunites. d) I''m much more interested in engineering excellence than mass market appeal. We''ve all seen the articles where someone declares that ZFS has upset the status quo and will cause the hi-end dedicated hardware RAID vendors major heartache. This will happen *only* if there is an entire/complete eco-system ready/willing/able to fill the gap between ZFS/host-based-RAID and the encumbent hardware RAID providers. And I''m talking about a vendor with the technical credibility of Sun.com and not some fly-by-night.storage.com outfit that no-one will trust. And then there is netapp.com ... which Sun/ZFS should declare a *target* of massive opportunity and completely obliterate. Sun needs to provide the necessary ammunition to capitalize on the "ZFS Advantage" and lead the market with high-bandwidth, direct attach, JBOD based storage sub-systems. Don''t ignore this ZFS based market opportunity and let the encumbent vendors continue to operate without the ZFS technology challenge that''ll take serious dollars away from their bottom line, if correctly deployed (as a commercial weapon of mass *migration*). PS: On a personal note, I''m looking at extending/expanding genunix.com and have a proposal in front of Sun management for storage hardware sponsorship. And I''ve specified cost-effective, ZFS technology based, JBOB storage using the 2530 and the (Sun certified) LsiLogic SAS HBA. Looking at ZFS and the current Sun product storage lineup, reminds me of the situation many, many years ago, when AT&T had Unix Sys V - but could never figure out how to turn that superior technology into a serious revenue stream..... I''d really hate to see history repeated... Regards, Al Hopper Logical Approach Inc, Plano, TX. al at logical-approach.com Voice: 972.379.2133 Fax: 972.379.2134 Timezone: US CDT OpenSolaris Governing Board (OGB) Member - Apr 2005 to Mar 2007 http://www.opensolaris.org/os/community/ogb/ogb_2005-2007/
On June 13, 2007 7:51:21 PM -0500 Al Hopper <al at logical-approach.com> wrote:> It seems wasteful to (determine the required part number and) order a > 2530 JBOD "expansion" shelf and then return it if it does not work out....It''s a 2501 I think.> Obviously this is a huge hole in Suns'' current storage product line. > There is a huge need for a ZFS saavy JBOD "box" with a high speed host > interconnect. The nearest alternative I''ve been able to dig up is the > Promise VTrak J300S - but I''d rather buy a Sun (or StorageTek) part than > a Promise Tech part. And, regardless of what the profit margin is, I''d > rather see Sun with that profit margin than Promise Tech.Get a 6140 expansion shelf. It is known to work. Probably near to the same $/GB as the 2530/2501 since the 2530 only takes SAS drives up to 300GB, instead of cheaper and larger capacity SATA drives. (Ignoring the fact that you need $$$ FC HBAs instead of $ SAS HBAs.) There''s clearly too much money to be made on the hardware still, to start making cheap effective JBODs available. You can pretty much forget about selling the RAID hardware at that point. Anyway I agree Sun should fill this hole, but the 2530 misses the mark. I''d like to see a chassis that takes 750GB/1TB SATA drives, with SAS host ports. And sell just the chassis, so I can skip the 100%+ drive markup. I guess I''m looking for a Promise J300s, but at twice the price (which is worth it to get better engineered hardware). Of course there is thumper, which is an excellent value but which has HA issues and a high buy-in price. ...> We''ve all seen the articles where someone declares that ZFS has upset the > status quo and will cause the hi-end dedicated hardware RAID vendors > major heartache. This will happen *only* if there is an entire/complete > eco-system ready/willing/able to fill the gap between ZFS/host-based-RAID > and the encumbent hardware RAID providers. And I''m talking about a > vendor with the technical credibility of Sun.com and not some > fly-by-night.storage.com outfit that no-one will trust. And then there is > netapp.com ... which Sun/ZFS should declare a *target* of massive > opportunity and completely obliterate.I don''t know, I rather like that Sun is providing all the ammunition for a Netapp competitor to start a war. Why hasn''t anyone done it? It''s still hard. And expensive. -frank
Hi folks, So the expansion unit for the 2500 series is the 2501. The back-end drive channels are SAS. Currently it is not "supported" to connect a 2501 directly to a SAS HBA. SATA drives are in the pipe, but will not be released until the RAID firmware for the 2500 series officially supports the SATA drives. The current firmware does not lock out those drives and prematurely releasing the drives would result in lots of service calls for unsupported configurations. The 750GB and 1TB drives are on the map behind the initial release of SATA support. The 2500 series engineering team is talking with the ZFS folks to understand the various aspects of delivering a complete solution. (There is a lot more to it than "it seems to work"...). -Joel This message posted from opensolaris.org
On June 26, 2007 2:13:54 PM -0700 Joel Miller <joel.miller at sun.com> wrote:> The 2500 series engineering team is talking with the ZFS folks to > understand the various aspects of delivering a complete solution. (There > is a lot more to it than "it seems to work"...).Great news, you made my day! Any ETA? -frank
Hello Joel,
Tuesday, June 26, 2007, 10:13:54 PM, you wrote:
JM> Hi folks,
JM> So the expansion unit for the 2500 series is the 2501.
JM> The back-end drive channels are SAS.
JM> Currently it is not "supported" to connect a 2501 directly to a
SAS HBA.
But does it work? Has anyone actually tested it? Any issues? Does
MPxIO work in such a config?
--
Best regards,
Robert mailto:rmilkowski at task.gda.pl
http://milek.blogspot.com
Hi Robert, It should work. We have not had the time or resources to test it (we are busy qualifying the 2530 (SAS array) with an upcoming MPxIO enabled MPT driver and SATA drive support). I do not know if MPxIO will claim raw drives or not....typically there are vendor specific modules that provide the identification and failover control...so I need to check. -Joel Robert Milkowski wrote:> Hello Joel, > > Tuesday, June 26, 2007, 10:13:54 PM, you wrote: > > JM> Hi folks, > > JM> So the expansion unit for the 2500 series is the 2501. > JM> The back-end drive channels are SAS. > > JM> Currently it is not "supported" to connect a 2501 directly to a SAS HBA. > > But does it work? Has anyone actually tested it? Any issues? Does > MPxIO work in such a config? > >
Hello Joel,
Wednesday, July 18, 2007, 8:19:56 PM, you wrote:
JM> Hi Robert,
JM> It should work. We have not had the time or resources to test it (we are
JM> busy qualifying the 2530 (SAS array) with an upcoming MPxIO enabled MPT
JM> driver and SATA drive support).
JM> I do not know if MPxIO will claim raw drives or not....typically there
JM> are vendor specific modules that provide the identification and failover
JM> control...so I need to check.
I was thinking about symmetric mode here...
--
Best regards,
Robert mailto:rmilkowski at task.gda.pl
http://milek.blogspot.com
On June 14, 2007 1:56:05 AM -0700 Frank Cusack <fcusack at fcusack.com> wrote:> Anyway I agree Sun should fill this hole, but the 2530 misses the mark. > I''d like to see a chassis that takes 750GB/1TB SATA drives, with SAS > host ports. And sell just the chassis, so I can skip the 100%+ drive > markup. I guess I''m looking for a Promise J300s, but at twice the > price (which is worth it to get better engineered hardware).I''m going to go ahead and give myself credit for the J4000 line of products. :) :) :)
On Wed, 16 Jul 2008, Frank Cusack wrote:> On June 14, 2007 1:56:05 AM -0700 Frank Cusack <fcusack at fcusack.com> wrote: > > Anyway I agree Sun should fill this hole, but the 2530 misses the mark. > > I''d like to see a chassis that takes 750GB/1TB SATA drives, with SAS > > host ports. And sell just the chassis, so I can skip the 100%+ drive > > markup. I guess I''m looking for a Promise J300s, but at twice the > > price (which is worth it to get better engineered hardware). > > I''m going to go ahead and give myself credit for the J4000 line of > products. :) :) :)*Ahem* http://richteer.blogspot.com/2007/04/close-but-no-cigar.html and http://richteer.blogspot.com/2006/05/sun-storage-product-i-would-like-to.html :-) -- Rich Teer, SCSA, SCNA, SCSECA CEO, My Online Home Inventory URLs: http://www.rite-group.com/rich http://www.linkedin.com/in/richteer http://www.myonlinehomeinventory.com
On July 16, 2008 9:40:03 PM -0700 Rich Teer <rich.teer at rite-group.com> wrote:>http://richteer.blogspot.com/2006/05/sun-storage-product-i-would-like-to.html I remember that! The 2.5" disks don''t really count as low cost, but still your other post beats me. :P Let''s say it was a team effort. :) -frank ps. kudos to Sun.