Hi, We have purchased a new Sun (Oracle) T3-2 machine and 5 shelves of 12 x 2TB SAS2 JBOD disks for our new backup server. Each shelf is connected via a single SAS cable to a seperate SAS controller. When the system arrived it had Solaris 10 U9 preinstalled. We tested ZFS performance and got roughly 1.3GB / sec when we configured a RADIZ2 per 12 disks, and roughly 1.7GB / sec when we configured a RAIDZ1 per 6 disks. Good performance. Then I noticed that some packages we''re missing from the installation, and I decided to jumpstart the server to get the server installed the same way as every other Solaris server we have. After it''s been reinstalled I get 100-300MB / sec, and very sporadic writes. I''ve played around with turning on and off mpxio on the mpt_sas driver, disabling increased the performance from 30MB / sec, but it''s still far from the original performance. I''ve attached some dumps of zpool iostat before and after reinstallation. Any suggestions to what settings might cause this? What to try to increase the performance? Regards, Siggi Before: pool0 51.5G 109T 93 10.4K 275K 1.27G pool0 59.8G 109T 92 10.8K 274K 1.32G pool0 68.1G 109T 92 10.6K 274K 1.30G pool0 76.5G 109T 92 11.4K 274K 1.39G pool0 85.0G 109T 92 10.2K 274K 1.25G pool0 93.6G 109T 92 11.2K 274K 1.37G pool0 93.6G 109T 0 9.77K 0 1.20G pool0 102G 109T 1 10.6K 5.99K 1.30G pool0 111G 109T 0 11.5K 0 1.41G pool0 119G 109T 1 10.7K 5.99K 1.31G pool0 127G 109T 89 11.1K 268K 1.36G pool0 136G 109T 1 11.8K 5.99K 1.44G After: pool0 30.8G 109T 0 297 0 36.2M pool0 30.8G 109T 0 2.85K 0 358M pool0 33.7G 109T 0 760 0 91.5M pool0 33.7G 109T 0 0 0 0 pool0 33.7G 109T 0 0 0 0 pool0 33.7G 109T 0 0 0 0 pool0 33.7G 109T 0 954 0 117M pool0 33.7G 109T 0 2.03K 0 255M pool0 36.2G 109T 0 358 0 42.2M pool0 36.2G 109T 0 0 0 0 pool0 36.2G 109T 0 858 0 105M pool0 36.2G 109T 0 1.28K 0 160M pool0 38.4G 109T 0 890 0 107M pool0 38.4G 109T 0 0 0 0 pool0 38.4G 109T 0 0 0 0 pool0 38.4G 109T 0 1.57K 0 197M pool0 40.3G 109T 0 850 0 103M pool0 40.3G 109T 0 0 0 0 pool0 40.3G 109T 0 0 0 0 pool0 40.3G 109T 0 0 0 0 pool0 40.3G 109T 0 2.58K 0 320M pool0 42.2G 109T 0 12 0 102K
sigbjorn at nixtra.com said:> I''ve played around with turning on and off mpxio on the mpt_sas driver, > disabling increased the performance from 30MB / sec, but it''s still far from > the original performance. I''ve attached some dumps of zpool iostat before and > after reinstallation.I find "zpool iostat" is less useful in telling what the drives are doing than "iostat -xn 1". In particular, the latter will give you an idea of how many operations are queued per drive, and how long it''s taking the drives to handle those operations, etc. On our Solaris-10 systems (U8 and U9), if mpxio is enabled, you really want to set loadbalance=none. The default (round-robin) makes some of our JBOD''s (Dell MD1200) go really slow for writes. I see you have tried with mpxio disabled, so your issue may be different. You don''t say what you''re doing to generate your test workload, but there are some workloads which will speed up a lot if the ZIL is disabled. Maybe that or some other /etc/system tweaks were in place on the original system. Also use "format -e" and its "write_cache" commands to see if the drives'' write caches are enabled or not. Regards, Marion
Hi, This turned out to be a scheduler issue. The system was still running the default TS scheduler. By switching to the FSS scheduler the performance was back to what it was before the system was reinstalled. When using the TS scheduler the writes would not evenly spread across the drives. We have in total 60 drives in the pool, and a few of the drives would peak in write throughput, while the other disks we''re almost idle. After changing to the FSS scheduler all writes we''re evenly distributed across the drives. This would just affect the user processes generating the load, as the zpool processes are still using the SDC scheduler introduced in S10 U9. I will do some testing on the loadbalance on/off. We have nearline SAS disks, which does have dual path from the disk, however it''s still just 7200rpm drives. Are you using SATA , SAS or SAS-nearline in your array? Do you have multiple SAS connections to your arrays, or do you use a single connection per array only? Rgds, Siggi On Wed, March 2, 2011 18:19, Marion Hakanson wrote:> sigbjorn at nixtra.com said: >> I''ve played around with turning on and off mpxio on the mpt_sas driver, >> disabling increased the performance from 30MB / sec, but it''s still far from the original >> performance. I''ve attached some dumps of zpool iostat before and after reinstallation. > > I find "zpool iostat" is less useful in telling what the drives are > doing than "iostat -xn 1". In particular, the latter will give you an idea of how many operations > are queued per drive, and how long it''s taking the drives to handle those operations, etc. > > On our Solaris-10 systems (U8 and U9), if mpxio is enabled, you really > want to set loadbalance=none. The default (round-robin) makes some of our JBOD''s (Dell MD1200) go > really slow for writes. I see you have tried with mpxio disabled, so your issue may be different. > > > You don''t say what you''re doing to generate your test workload, but there > are some workloads which will speed up a lot if the ZIL is disabled. Maybe that or some other > /etc/system tweaks were in place on the original system. > Also use "format -e" and its "write_cache" commands to see if the drives'' > write caches are enabled or not. > > Regards, > > > Marion > > > >
sigbjorn at nixtra.com said:> I will do some testing on the loadbalance on/off. We have nearline SAS disks, > which does have dual path from the disk, however it''s still just 7200rpm > drives. > > Are you using SATA , SAS or SAS-nearline in your array? Do you have multiple > SAS connections to your arrays, or do you use a single connection per array > only?We have four Dell MD1200''s connected to three Solaris-10 systems. Three of the MD1200''s have nearline-SAS 2TB 7200RPM drives, and one has SAS 300GB 15000RPM drives. All the MD1200''s are connected with dual SAS modules to a dual-port HBA on their respective servers (one setup is with two MD1200''s daisy-chained, but again using dual SAS modules & cables). Both types of drives suffer super-slow writes (but reasonable reads) when loadbalance=roundrobin is in effect. E.g 280 MB/sec sequential reads, and 28MB/sec sequential writes, for the 15kRPM SAS drives I tested last week. We don''t see this extreme slowness on our dual-path Sun J4000 JBOD''s, but those all have SATA drives (with the dual-port interposers inside the drive sleds). Regards, Marion