raid controller raid 1 on 4 drives

greengraph

Reputable
Aug 16, 2014
7
0
4,510
I need a high read speed which can possible have from 4 SSD array(RAID).
I chose a RAID 1 but my raid controller disallow that, only 2 drives. 4 drives only RAID 0 or 10.
RAID 10 - 4 drives just double drive capacity(I don't need it!) but there is no read speed acceleration in comparison with RAID 1 - 2 drives...
Is there a hardware raid card which allow 4 drives for RAID 1?
 
Why would you want to mirror 4 disks? That sounds like overkill (and it's certainly not going to be faster than a single drive). The chances of 4 disks all failing at once are so small that the manufacturers just don't let you so something that would, frankly, be nonsensical.

For speed you want RAID 0 (at the expense of reliablity), but I'm not convinced that 4 SSDs in a RAID 1 array will be that much faster than a good, larger SSD.
 
You can't get a high read/write speed to 4 mirrored disks as you are creating 4 time the work. RAIDing SSD's in a stripped set has no real world gains anyhow and you run the risk of losing everything on the array as if one disk goes everything does. If you must use all 4 as you need the space in one partition just JBOD 2 and mirror it with the other two which is RAID 1+0=RAID 10
 
Parkdale software test:
RAID/SSD - read/write speed mbps
no/1 - 200/120
1/2 - 300/120
10/4 - 350/150
0/2 - 350/200
0/4 - 350/300

Declared interfaces speed:
SATA 1 - 1,5 gbps
SATA 2 - 3 gbps
PCIE 1 - 2,5 gbps
PCIE 2 - 5 gbps

What for those speeds if SSD(which speed declared by manufacturer is 450 mbps in both direction) in real just 200/120 mbps and RAID with 4 SSD maximum just 350/300 mbps???

PS. For me the data survival is second because it's small and changes very slow, loss of some data not critical. I have an image of disk and can rebuild OS for a 10 minutes. Other program make a daily archive of documents to a separate HD.
 
I suspect that you are confusing RAID 1 and RAID 0. RAID 1 is mirroring, which means that you will only get 1/4 of the storage capacity out of your disks, with little speed advantage, and reliability beyond all reason.

RAID 0 would give you a performance increase by striping the data across multiple disks, but I would still investigate the performance differences between small and large SSDs. This can be a bigger factor than the advantages of striping.
 
Ijack,
Confusing? I knew that RAID 1 - mirroring, RAID 0 - striping, RAID 10 is 1+0.
Speed tests I published.
I want to test RAID 1 with 4 SSD and my RAID card disallow that!
In theory, as I understand, RAID 1 read speed can be faster than RAID 0 because identical data read from 4 drives in RAID 1 but in RAID 0 there is no identical data at all!
 
Fair enough. I thought you must be confused because you seem to think that you will get better performance from RAID 1 than RAID 0 or RAID 10. As you understand RAID you obviously know that this is not the case. For some reason you desire the moderate performance of RAID 1 rather than the faster RAID 10 (or 0); that's your choice to make.

In theory, as I understand, RAID 1 read speed can be faster than RAID 0
This is quite wrong. RAID 0 or RAID 10 will always provide faster performance than RAID 1. And, as I have said before, large SSDs normally provide much enhanced performance over smaller ones or RAID arrays of smaller ones. As you have discovered, mirroring 4 disks is almost unheard of, to the extent that your controller doesn't allow it, because it provides no performance advantage at the expense of wasting most of your storage space.

But do what you like; even though you would get better performance, twice the capacity, and enhanced reliability with a RAID 10 array, you may be able to find an add-on controller that would allow you to set up a RAID 1 array with 4 disks.
 
Ijack,
Theory and practice are not the same. In theory Kingston SSD have a speed of 450 to 555 mbps. In practice - maximum of 200 mbps. And even RAID with 4 SSD not reaches the declared speed for 1.
In theory, there are no physical limitations(as you wrote about what I can do ...) to create a hardware RAID 1 with 4 discs, in practice there is - because I can't find RAID controller which can do it(So I can't do it even if I want to)...
 
I'll leave you to it. You've published tests showing that what you want to do is going to lead to a slower system. If that's what you want it's not my place to continue to try to dissuade you.
 

Anyway, Thank you for a thoughts direction! I'll try to find better SSD.
In my example, it was Kingston SSDnow V300 120Gb

 

Theory & practice ... Prove it!
 
Stripping two SSD's doesn't give any real world gains. I didn't believe it too when I was told by someone and had to research it when install my two SSD's.

"If you're planning an upgrade and want to know whether to buy a couple of 128 GB drives and put them in RAID 0 or just grab a single 256 GB SSD, for example, the answer still seems clear enough to us: just grab the large drive and use one..."
http://www.tomshardware.com/reviews/ssd-raid-benchmark,3485-13.html





 
I think it varies depending upon the SSDs in question and whether they are SATA 2 or SATA 3. There have been several tests which do show that there is a marginal increase in speed when using two SATA 3 SSDs in a RAID 0 setup. To my mind this is not worth the risks involved and can be bettered by just using a single SSD with double the capacity of the smaller ones.

If you really want the best performance I think the solution is to use an SSD that plugs into the PCIe bus, rather than SATA, but they tend to be expensive.
 
I read a lot of messages around the WWW about the different manufacturers SSDs high death rate. So I stopped at the Raid 10 and test this SSDs death theory.
From this point of view 1 large SSD is equal to RAID 0 with small SSDs.
I also read an opinion that large SSDs high speed is a fake - depend on larger cache, most of tests represent how data is temporarily stored in cache with higher speed but the real data store speed is much slow.

I also saw some tests inaccuracy when changes the RAID stripe size to smaller one - there is no significant difference(about 2% deceleration) in test but in specific software(not testing software) with specific operation I received about 40% of acceleration.


This is why I want to wait a little to see how SSDs will be in time.