Using glance plus, I see that one or two of my disks are maxing out at 100% I/O. These are in a va7410 array, RAID 1+0. How can I spread this across additional disks to improve performance. Is there any tool/utility to see how the striping is actually laid out?
Tek-Tips is the largest IT community on the Internet today!
Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!
-
Congratulations strongm on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!
Disk I/O Bottleneck 1
- Thread starter greglocke
- Start date
-
1
- #2
You can either use lvm to stripe or more manually, extent based striping.
lvm will stripe across disks using to methods ...
-i (lvm storts out the stripes)
-D (distributed extent based striping using physical volume groups)
Persoanlly, if possible, I'd blow away the data and then recreate using extent based striping, if you have alternative paths to each physical disk ...
eg. We use mainly XP array or EMC, everything is dual pathed, so for each device, we have an alternate path for resiliance, meaning two HBAs in each server, so if performance is an issue, we can use extent based striping, and by controlling which devices we alternate round, one extent goes thru. one HBA, the next extent thru. the other HBA, so distribute the load across two HBAs. I've seen some amazing performance gains using this (knocked down from 100% to 50%ish).
Martin
Martin
lvm will stripe across disks using to methods ...
-i (lvm storts out the stripes)
-D (distributed extent based striping using physical volume groups)
Persoanlly, if possible, I'd blow away the data and then recreate using extent based striping, if you have alternative paths to each physical disk ...
eg. We use mainly XP array or EMC, everything is dual pathed, so for each device, we have an alternate path for resiliance, meaning two HBAs in each server, so if performance is an issue, we can use extent based striping, and by controlling which devices we alternate round, one extent goes thru. one HBA, the next extent thru. the other HBA, so distribute the load across two HBAs. I've seen some amazing performance gains using this (knocked down from 100% to 50%ish).
Martin
Martin
- Status
Similar threads
- Locked
- Question