up vote 0 down vote favorite
share [g+] share [fb]

I build a computer with a raid5, using the motherboards raid controller (ASUS P7H57D-V EVO - intel Rapid Storage Technology). The read and write are however very slow, when using the raid controller :( - I am using Windows 2008 R2, and when using the windows software raid, it was ok in speed - so there must be an issue with the controller? Im using 6 disks on 2TB each.

Do anyone have any idea why its so slow, and how to fix it? I rather not pick the easy solutiuon of "just buy a raid controller" :|

If you need more info about my setup, please just ask.

Thanks :)

link|improve this question
25% accept rate
2  
Not a solution, but you really shouldn't be using RAID-5 with multi-terabyte drives. You're quite likely to have a failure during rebuild, see zdnet.com/blog/storage/why-raid-5-stops-working-in-2009/162. With drives as cheap as they are, using RAID-10 is far safer (and performs much better for writes). – rmalayter Sep 16 '10 at 20:44
@rmalayter, note that most newer drives are a couple orders of magnitude less likely to have an error; which is not accounted for in the linked article. URE of 10^16 is pretty common, and 10^17 is being seen in common SATA drives. I've seen enterprise SAS drives quoted as high as 10^21. – Chris S♦ Jul 26 '11 at 1:46
Very bad question. What is "very slow"? (rhetoric question) – poige Jul 26 '11 at 6:09
@Chirs S, the Seagate barracuda XT and Western Digital Caviar Black are still rated at URE=10^14. These are the mainstream multi-terabyte "volume" SATA drives from each vendor. RAID5=bad on such drives (assuming the UREs from the manufacturer can be trusted). Of course "enterprise near-line" drives might be better (even if they use the same mechanisms, they might have more ECC in the firmware layer). – rmalayter Aug 1 '11 at 14:21
This is why its recommended on linux to get MDADM to run a badblock check every few weeks. echo check >> /sys/block/mdX/md/sync_action – Silverfire Sep 27 '11 at 1:53
show 1 more comment
feedback

2 Answers

up vote 1 down vote

Did you just build the array? its possible that it is still syncing all the drives. on a 5x1TB array it took me 9 hours and during that time it was very very slow as there was a lot of background processing and writes.

link|improve this answer
Given no information from the OP on the state of the array syncing going on ins a likely culprit. It takes ages of time to sync a large raid and during this time - obviously - a lot of the IO budget is tied up for the sync process. – TomTom Dec 29 '11 at 8:52
feedback
up vote 0 down vote

Check driver versions. It should not be high performance compared to a controller, but it should be comparable to windows own RAND implementation.

link|improve this answer
I did install the latest version of the driver - no change :( – Cederstrom Sep 16 '10 at 21:17
feedback

Your Answer

 
or
required, but never shown
discard

Not the answer you're looking for? Browse other questions tagged or ask your own question.