topic Re: MSA1000 big performance issue in MSA Storage https://community.hpe.com/t5/msa-storage/msa1000-big-performance-issue/m-p/694232#M2522 You get what you pay for. The MSA1000 is not a high performance based Array. HP says optimimum disk configuration is using 7 or more disks. Array disks should be in slots 0-7 or 8-14. Preferably don't span across these sets if you are configuring for performance. I did testing and got similar results. SCSI will always out perform this array. Remember: the Contrlr is only 160mb/s. 15K drives would also help. Good luck Tue, 05 Sep 2006 08:23:17 GMT James Muell 2006-09-05T08:23:17Z MSA1000 big performance issue https://community.hpe.com/t5/msa-storage/msa1000-big-performance-issue/m-p/694231#M2521 Hello,<BR /> i'm using a MSA1000 SAN with 4 * 10k 144GB U320 Disks in a Raid 0+1 config on a debian/sarge with custom 2.6.17.8 Kernel. System was a BL20pG2 Blade.<BR /> <BR /> dmesg spits out:<BR /> QLogic Fibre Channel HBA Driver<BR /> ACPI: PCI Interrupt 0000:01:03.0<A rel="nofollow" target="_blank"> -&gt; GSI 23 (level, low) -&gt; IRQ 201<BR /> qla2300 0000:01:03.0: Found an ISP2312, irq 201, iobase 0xf8838000<BR /> qla2300 0000:01:03.0: Configuring PCI space...<BR /> qla2300 0000:01:03.0: Configure NVRAM parameters...<BR /> qla2300 0000:01:03.0: Verifying loaded RISC code...<BR /> qla2300 0000:01:03.0: Extended memory detected (512 KB)...<BR /> qla2300 0000:01:03.0: Resizing request queue depth (2048 -&gt; 4096)...<BR /> qla2300 0000:01:03.0: Allocated (1308 KB) for firmware dump...<BR /> qla2300 0000:01:03.0: Waiting for LIP to complete...<BR /> qla2300 0000:01:03.0: Cable is unplugged...<BR /> scsi0 : qla2xxx<BR /> qla2300 0000:01:03.0:<BR /> QLogic Fibre Channel HBA Driver: 8.01.06-fo<BR /> QLogic QLA2344 - 133MHz PCI-X to 2Gb FC, Quad Channel<BR /> ISP2312: PCI-X (100 MHz) @ 0000:01:03.0 hdma-, host#=0, fw=3.03.19 IPX<BR /> ACPI: PCI Interrupt 0000:01:03.1<B> -&gt; GSI 18 (level, low) -&gt; IRQ 209<BR /> qla2300 0000:01:03.1: Found an ISP2312, irq 209, iobase 0xf883a000<BR /> qla2300 0000:01:03.1: Configuring PCI space...<BR /> qla2300 0000:01:03.1: Configure NVRAM parameters...<BR /> qla2300 0000:01:03.1: Verifying loaded RISC code...<BR /> qla2300 0000:01:03.1: Extended memory detected (512 KB)...<BR /> qla2300 0000:01:03.1: Resizing request queue depth (2048 -&gt; 4096)...<BR /> qla2300 0000:01:03.1: Allocated (1308 KB) for firmware dump...<BR /> qla2300 0000:01:03.1: Waiting for LIP to complete...<BR /> qla2300 0000:01:03.1: LIP reset occured (f8f7).<BR /> qla2300 0000:01:03.1: LIP occured (f8f7).<BR /> qla2300 0000:01:03.1: LOOP UP detected (2 Gbps).<BR /> qla2300 0000:01:03.1: LIP reset occured (f7f7).<BR /> qla2300 0000:01:03.1: LIP occured (f7f7).<BR /> qla2300 0000:01:03.1: Topology - (Loop), Host Loop address 0x0<BR /> qla2300 0000:01:03.1: Waiting for LIP to complete...<BR /> qla2300 0000:01:03.1: Topology - (Loop), Host Loop address 0x0<BR /> scsi(1) :Loop id 0x007d is a MSA1000 device<BR /> scsi(1) :Loop id 0x007d is a MSA1000 device<BR /> qla2x00_combine_by_lunid: Found a controller path 0x0 - lun 0<BR /> qla2x00_combine_by_lunid: Found a controller path 0x0 - lun 0<BR /> scsi1 : qla2xxx<BR /> qla2300 0000:01:03.1:<BR /> QLogic Fibre Channel HBA Driver: 8.01.06-fo<BR /> QLogic QLA2344 - 133MHz PCI-X to 2Gb FC, Quad Channel<BR /> ISP2312: PCI-X (100 MHz) @ 0000:01:03.1 hdma-, host#=1, fw=3.03.19 IPX<BR /> Vendor: COMPAQ Model: MSA1000 Rev: 4.24<BR /> Type: RAID ANSI SCSI revision: 04<BR /> qla2300 0000:01:03.1: scsi(1:0:0:0): Enabled tagged queuing, queue depth 32.<BR /> 1:0:0:0: Attached scsi generic sg0 type 12<BR /> Vendor: COMPAQ Model: MSA1000 VOLUME Rev: 4.24<BR /> Type: Direct-Access ANSI SCSI revision: 04<BR /> qla2300 0000:01:03.1: scsi(1:0:0:1): Enabled tagged queuing, queue depth 32.<BR /> SCSI device sda: 573496752 512-byte hdwr sectors (293630 MB)<BR /> sda: Write Protect is off<BR /> sda: Mode Sense: 83 00 00 08<BR /> SCSI device sda: drive cache: write back<BR /> SCSI device sda: 573496752 512-byte hdwr sectors (293630 MB)<BR /> sda: Write Protect is off<BR /> sda: Mode Sense: 83 00 00 08<BR /> SCSI device sda: drive cache: write back<BR /> sda: sda1<BR /> sd 1:0:0:1: Attached scsi disk sda<BR /> sd 1:0:0:1: Attached scsi generic sg1 type 0<BR /> <BR /> <BR /> <BR /> <BR /> I'm using the qlogic driver taken from qlogic ftp site (not the included one inside vanilla kernel, cause they don't work with ACU):<BR /> <BR /> version: 8.01.06<BR /> <BR /> <BR /> With "dd if=testfile_4gb of=/dev/null" i only got &lt;30MB/s. <BR /> <BR /> Local Harddisk (72GB 10k @ Smart Array 5i/532 Controller) results in ~70MB/s.<BR /> <BR /> Some tests with Raid0 on MSA1000 performs with ~160MB write and 80MB/s on read (yes _faster_ write speed than read) a 4GB testfile.<BR /> <BR /> Looks really curious to mee, anybody seen something similar and has a solution? MSA1000 with 30MB/s in Raid0+1 is useless for us.<BR /></B></A> Mon, 04 Sep 2006 07:18:36 GMT https://community.hpe.com/t5/msa-storage/msa1000-big-performance-issue/m-p/694231#M2521 nias 2006-09-04T07:18:36Z Re: MSA1000 big performance issue https://community.hpe.com/t5/msa-storage/msa1000-big-performance-issue/m-p/694232#M2522 You get what you pay for. The MSA1000 is not a high performance based Array. HP says optimimum disk configuration is using 7 or more disks. Array disks should be in slots 0-7 or 8-14. Preferably don't span across these sets if you are configuring for performance. I did testing and got similar results. SCSI will always out perform this array. Remember: the Contrlr is only 160mb/s. 15K drives would also help. Good luck Tue, 05 Sep 2006 08:23:17 GMT https://community.hpe.com/t5/msa-storage/msa1000-big-performance-issue/m-p/694232#M2522 James Muell 2006-09-05T08:23:17Z