Hi
We are in the process of setting up a new Gluster setup which is going to be use by local Atlas users. I am looking into various raid and xfs options. I did some benchmarking and arrived on following options
Six Dell Power Edge server , Twelve 8TB disk per server with Raid 6 . Initial size of the file system will be around 480 TB
RAID setting
ID : 1
Status : Ok
Name : sdb
State : Ready
Hot Spare Policy violated : Not Assigned
Encrypted : No
Layout : RAID-6
Size : 74,415.00 GB (79902497832960 bytes)
T10 Protection Information Status : No
Associated Fluid Cache State : Not Applicable
Device Name : /dev/sdb
Bus Protocol : SAS
Media : HDD
Read Policy : Read Ahead
Write Policy : Write Back
Cache Policy : Not Applicable
Stripe Element Size : 256 KB
Disk Cache Policy : Unchanged
I am confused between "Read Ahead" and "Adaptive Read Ahead". What is the preferred option? Is 256 KB stripe element size OK or I should go for 512KB?
I used these options for mkfs
mkfs.xfs -f -d su=256k,sw=10 /dev/sdb
It gave better iozone results than default options. I want to pass this through you guys to make sure that I am not doing anything completely wrong. Please feel free to suggest any other optimization.
Thanks
Kashif
|