Take the 2-minute tour ×
Super User is a question and answer site for computer enthusiasts and power users. It's 100% free, no registration required.

I have a quick question regarding what I'm seeing as an obstacle in getting the proper performance out of a RAID card...

I'm interested in setting up two RAID arrays (a RAID0 of 2 disks, and a RAID1 of another 2 disks), and am looking at a hardware RAID solution. I am interested in the LSI 9260-4i in particular.

Now my question is this: this RAID card is advertised as having 6Gbps SATA support. What I don't get though is that as a PCI-E 2.0 x8 card, it should theoretically top-off at 8 Gbps - so I don't get how it can support 4 (or, depending on the model, 8) channels of 6Gbps drives.

Basically, I want to create a 2x RAID0 array of 6Gbps drives (OCZ Vertex 3 240GB), and a 2x RAID1 array of 6Gbps drives (WD Velociraptor 600GB). I realize that neither of these disks takes full advantage of the 6.0Gbps SATA III interface, but they do exceed the 3.0Gbps SATA II interface specifications by a nice amount even in standalone configurations.

So I don't understand how this card can support ~8Gbps (2x SSD in RAID0) + 4Gbps (2x WD in RAID1) without reaching the limits of the underlying interface.

Are there any PCI 2.0 x16 RAID cards available, because I couldn't find any... And I'm really not looking forward to buying two of these cards to pull this off, but if that's the only choice...

share|improve this question

2 Answers 2

up vote 1 down vote accepted

Found the answer, thanks to some insight on another online forum.

In my original post, I was mistakenly said the RAID0 would be taking up ~8Gbps during burst speeds which is incorrect. The physical sectors are being written at around 8Gbps maximum speed, but the data being transferred (the logical sectors) is only half that at around 4Gbps.

So now we're looking at a maximum (and unlikely) 4Gbps burst speed on both arrays totaling at 8Gbps, which is perfectly fine for an 8-lane PCI-E 2.0 interface.

share|improve this answer

It's because you aren't transferring data all at one time. They break it so you can do things like error correction.

They just queue it up and move your data through. It's like moving to another city. No one thinks "OMG, you fit your WHOLE HOUSE in that moving van?" But rather they realized that you boxed it up and even possibly made more than one trip. . .

share|improve this answer
    
What? That really has nothing to do with my question. To play along withy your metaphor, I have two moving vans of total size x (12Gbps) that I'm trying to fit into one large moving van of size y (8Gbps), and I'm wondering how that can be possible without making more trips to send the same data. –  Mahmoud Al-Qudsi Aug 2 '11 at 18:25
    
The point is they can say they support 4 or 6 SATA 3.0 ports is because it's theoretical. On top of the overhead of sata, there are other components in the system. If your application isn't ready to accept any data or the CPU is busy processing something else, then your effective bandwidth is zero. Yet, data isn't being corrupted nor packets being dropped. Why? Because your card knows that it isn't the only component in the system and thus acts like a well behaved piece of hardware. –  surfasb Aug 3 '11 at 0:52
    
It isn't like you caught them in a false marketing scheme or some red handed conspiracy. It's because there is more to motherboard design than simply shoving boxes down a chute. And thus these numbers you read are just that. They hide the complexity from you. It isn't a perfect abstraction as you have noticed, but it doesn't mean there isn't any truth to it. And that's how it's done. –  surfasb Aug 3 '11 at 0:53
    
And we haven't even gotten started on PCIe packet overhead. . . –  surfasb Aug 3 '11 at 1:00

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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