You will see the same performance on S10 overriding the default behaviour that you saw while running S8/SDS 4.2.1. I have looked at the bug report associated with this bug and it states that some baselining was done to determine whether this fix was worthwhile. It apparently was since the fix was putback but the amount of performance gain is not listed in that report. Let me talk to the person that made the fix to see what the approximate gain was. If the information is available I will post it here.
Harold prasad jlv wrote: >Harold: > >Thank you for that explanation. > >Are we going to experience performance problems (& how >significant) if we use "metainit -o" to overwrite the >default behaviour? > >-- prasad > >--- Harold Shaw <Harold.Shaw at Sun.COM> wrote: > > > >>The reason that the default has changed is that a >>bugfix went into >>Solaris 10 that was later backported into Solaris 9 >>that made the >>default soft partition creation cylinder aware. It >>was noticed that >>soft partitions that were created on top of stripes >>could incur >>additional physical disk IOs if they weren't aligned >>properly. This can >>be manually overridden if desired by using the '-o' >>parameter of metainit. >> >>Harold Shaw >> >>prasad wrote: >> >> >> >>>Here'a little more information about the soft >>> >>> >>parition difference between SDS 4.2.1 on Solaris 8 >>and SVM in Solaris 10: >> >> >>>Solaris_8_202# metastat -p d350 >>>d350 -p d520 -o 10177 -b 2097152 >>>d520 -m d420 1 >>>d420 1 4 c1t2d0s0 c1t3d0s0 c1t4d0s0 c1t5d0s0 -i 32b >>> >>>Solaris_10_FCS# metastat -p d350 >>>d350 -p d520 -o 10208 -b 2097152 >>>d520 -m d720 1 >>>d720 1 4 c2t10d0s0 c2t11d0s0 c2t12d0s0 c2t13d0s0 -i >>> >>> >>32b >> >> >>>where d350 is the soft parition that is created >>> >>> >>using "metainit d350 -p 520 1g" in both releases. >> >> >>>The starting block is 10177 in Solaris 8 and 10208 >>> >>> >>in Solaris 10. Why is the starting block different? >>Is the soft parition storing more header info in >>Solaris 10? >> >> >>>-- prasad >>>This message posted from opensolaris.org >>>_______________________________________________ >>>lvm-discuss mailing list >>>lvm-discuss at opensolaris.org >>> >>> >>> >>> >> >> > > > > >__________________________________ >Yahoo! Mail - PC Magazine Editors' Choice 2005 >http://mail.yahoo.com > > -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://mail.opensolaris.org/pipermail/lvm-discuss/attachments/20050915/800273e9/attachment.html>
