• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Help me find the right RAID controller 8)

S

saqmaster

n00b
Joined
Aug 9, 2004
Messages
42
Hi all..

I have a mission.. if you choose to accept, it may not explode after 30 seconds..

I'm looking for a SATA RAID controller card. It's to go in a 33mhz 64-bit PCI slot (long slot, not pci-x). Out of interest, the server is a HP Netserver E800.

It needs to do hardware RAID 5. It'd be nice if it had some cache memory. It needs to support 4 SATA drives. SATA 2 support isn't important.

So far i've not managed to find anything that uses this slot. Only 32-bit PCI or PCI-X. Or 66mhz bus speeds.

Any ideas anyone? :cool:

Thanks, Stu.
 
saqmaster said:
33mhz 64-bit PCI slot (long slot, not pci-x)
Click to expand...

That is PCI-X but not PCI-Express.

PCI-X = 64bit PCI, an old standard from the Pentium 2 days.

PCI-Express = multilane serial standard, of recent design.

So you do want a PCI-X card. (but not a PCI-Express card)

==>Lazn
 
The Promise 4-Port SATA RAID controller does what you want.

It's running perfectly in my Media Server in a 32-bit PCI slot. It has 64MB cache standard and does hardware RAID 5.
 
Lazn_Work said:
That is PCI-X but not PCI-Express.

PCI-X = 64bit PCI, an old standard from the Pentium 2 days.

PCI-Express = multilane serial standard, of recent design.

So you do want a PCI-X card. (but not a PCI-Express card)

==>Lazn
Click to expand...

Actually, PCI-X is a faster/later standard. Here's some info taken from wikipedia :-

Wikipedia said:
PCI-X changes the protocol slightly and increases the data rate to 133MHz (peak transfer rate of 1066 MB/s)
PCI-X 2.0 specifies a 266MHz rate (peak transfer rate of 2133 MB/s) and also 533MHz rate, expands the configuration space to 4096 bytes, adds a 16-bit bus variant and allows for 1.5 volt signalling
Click to expand...

So as originally said, I have a 33mhz 64-bit PCI connector, not pci-x :cool:
 
valve1138 said:
The Promise 4-Port SATA RAID controller does what you want.

It's running perfectly in my Media Server in a 32-bit PCI slot. It has 64MB cache standard and does hardware RAID 5.
Click to expand...

I really want a 64-bit bus. 33mhz 32-bit PCI theoretical max is 130MB/sec. A good raid controller with fast disks can saturate that bus, and that's without considering the gigabit network card that's on the same bus trying to shift the data around! Basically i'd want a good 240MB/sec of bus bandwidth.. Which is where the 64-bit bus comes into it!
 
saqmaster said:
A useful document, thanks. I think technically what you're saying is PCI-X is backwards compatible, so will work. What I said was my slot wasn't a PCI-X one :cool:

Well, that makes life a bit easier.. i'll continue the hunt..
Click to expand...

Technically 64bit PCI was not called PCI-X untill it was running at 100mhz (PCI-Xa).

But really, PCI-X can be and often is used to describe 64bit PCI slots of all speeds.

==>Lazn
 
"Real" hardware raid is expensive. The only manufacturer I'd put the extra money on right now is Areca, and that's a *lot* of extra money. My favorite review shows the Highpoint holding its own against more expensive controllers. Worth a look, anyways.

 
unhappy_mage said:
"Real" hardware raid is expensive. The only manufacturer I'd put the extra money on right now is Areca, and that's a *lot* of extra money. My favorite review shows the Highpoint holding its own against more expensive controllers. Worth a look, anyways.

Click to expand...

Will, are you sure Tweaker's isn't paying you?

If not then they should!!


 
I'm all for it ^_^ As soon as somebody does a better job of it, or a similar job on the next generation of raid cards, I'll be the stuck-record proponent of their review ;)

 
Okay, I committed!

I purchased four WD 250GB 'Raid Edition' Disks and a Highpoint 1640 controller. Now, I know this isn't the best controller in the world, but after looking at some RAID 5 benchmarks against other cards (eg, Adaptec and Intel) in the ~$300 area (3x the price), it appears to be a better card. In some tests with 4 disks, it sustained 70MB/sec sequential read speed, which is pretty acceptable to me.

One misleading thing though is the Highpoint website doesn't say it supports RAID 5 but the product manual does, it also shows how to configure it. And some websites, including reviews say it is "software raid 5", which is a bit misleading. What I believe they mean is it doing have an additional processor on the card to offload the XOR functions so it either uses the microcontroller/chipset on the card, or perhaps the host machines CPU. Either way, it's not 'software raid' as most people know it (eg, configure it within the operating system). You configure the RAID 5 on this card within it's BIOS, so host o/s is irrelevant.

That's how I see it anyway, i'll let you know if it's anything otherwise once i've installed it :cool:

If anyone else has used one of these cards at RAID 5 level, let me know :D
 
Actually, a change of plan. I found some tests of the RR1640 which demonstrated it using 100% CPU on a 2.8 Ghz P4 whilst doing some heavy work. Considering the box this is going in is only a dual 733mhz, I deemed it unsuitable.

So the 1640 is being returned and i've ordered the RR1820A instead which has an onboard XOR processor.. It's tempting to buy another 4 disks too but I just dont have the enclosure or psu for them! :cool:
 
saqmaster said:
Actually, a change of plan. I found some tests of the RR1640 which demonstrated it using 100% CPU on a 2.8 Ghz P4
Click to expand...

That is why it is software RAID. It is setup in the BIOS of the card, but it the XOR is still done by the host CPU, thus software RAID.

Hardware RAID applies to a dedicated XOR processor.

==>Lazn
 
Lazn_Work said:
That is why it is software RAID. It is setup in the BIOS of the card, but it the XOR is still done by the host CPU, thus software RAID.

Hardware RAID applies to a dedicated XOR processor.

==>Lazn
Click to expand...
This means there's a badly written driver, probably that waits on IO. CPUs these days are *plenty* fast to do software raid; only a lousy driver (as apparently these have) would spend 100% of the cpu time waiting on blocks to come back from the disk. If you use Linux Software Raid (or probably the Windows equivalents) it won't use nearly as much CPU. Writing to a raid 5 at ~40 MB/s only takes 2% CPU on my machine.

 
unhappy_mage said:
This means there's a badly written driver, probably that waits on IO. CPUs these days are *plenty* fast to do software raid; only a lousy driver (as apparently these have) would spend 100% of the cpu time waiting on blocks to come back from the disk. If you use Linux Software Raid (or probably the Windows equivalents) it won't use nearly as much CPU. Writing to a raid 5 at ~40 MB/s only takes 2% CPU on my machine.

Click to expand...

That too. I seem to remember software RAID 5 in linux using something like 40% of a 450mhz Celeron's CPU.. So basically nothing on a modern CPU.

==>Lazn
 
They must have made it more efficient since then; this machine is only a dual p3-933, and it's only using 2% of one cpu.

 
So I built the server out with the 4x250gb HD's and the 1820A array controller.. Took about 2 hours 'ish to build the array..

It's pretty fast. HDTach reports 115MB/sec average read speed.. not bad for a fairly budget raid 5 card.. I need to do some fair comparisons on CPU though as the only data I have right now (~40-50% CPU) is whilst pumping ~150mbit/sec through the gigabit NIC at the same time.. :cool:
 
You must log in or register to reply here.
Back
Top