Recently I established a storage box and install the Open-E DSS V6 trial edition. The storage box was composed of:
1*Intel Xeon E5504 CPU,
2*2GB DDR3-1066 ECC+REG SDRAM,
1*Adaptec 5805 RAID controller with 8*Seagate Cheetah 15K.7 300GB SAS derviers, and these drivers were configured in a RAID 5 array,
1*Qlogic QLE2562 dual-port 8Gb FC HBA.
On this box I created 2 LUs, one for OS installation, while the other for performance test.
Another server was connected to the created LUs with a Qlogic QLE2560 single-port 8Gb FC HBA. They were connected with the Brocade 300 switch. Windows Server 2008 R2 Enterprise was installed to the first LU. Then I tested the transfer rate of the second LU by HDTune. The average rate is only about 120MB/s. I copied a file of 2GB from the first LU to the second LU, and the speed shown is also about 100MB/s.
My question is whether this is the real performance of this system, or should I do something to optomize the storage system?
Thanks!
Thank you, enealDC.
Because the storage would be used in the enviroment of Windows Servers, the performance under Windows is important to me. Thank you for your advice, I would do some tests under Linux.
No problem. Just to be clear, test the performance of the disks through the array (e.g. locally versus through the FC target) so that way you understand the true performance of your array on Linux, which Open-E is based on. This way, you can put your FC performance in it's true perspective.
In other words, don'' expect 1GB p/s through FC if you can only deliver 400MB p/s locally