Help me choose a good motherboard for NAS

[hasu]

I am RMA'ing a bad ASRock E3C224D4I-14S that pushed me back to square one and now shopping for a motherboard with at least 10 SATA ports. I like the following two options. What would be more appropriate option for a NAS? Will there be any difference in power consumption between the two?

1. The same ASRock with built-in controller card

ASRock E3C224D4I-14S (Newegg: N82E16813157486)
http://www.newegg.com/Product/Product.aspx?Item=N82E16813157486

2. Supermicro motherboard with a separate add-on SAS card

SUPERMICRO AOC-SAS2LP-MV8 (Newegg: N82E16816101792)
http://www.newegg.com/Product/Product.aspx?Item=N82E16816101792

SUPERMICRO MBD-X10SLM+-F-O (Newegg: N82E16813182823)
http://www.newegg.com/Product/Product.aspx?Item=N82E16813182823

 

[dave_the_nerd]

Well, I have the second one, so my vote's biased.

Also, I'm cheap. If you skip the SAS card with the Supermicro and just use an couple el-cheapo SATA cards (a 4-port is usually $20-$30 each) you would save some bucks.

Even more if you wait for Newegg to have an open-box X10SLM. I only paid $130 for mine.

Note that the Supermicro also requires ECC RAM.

 

[hasu]

Well, I have the second one, so my vote's biased.

Also, I'm cheap. If you skip the SAS card with the Supermicro and just use an couple el-cheapo SATA cards (a 4-port is usually $20-$30 each) you would save some bucks.

Even more if you wait for Newegg to have an open-box X10SLM. I only paid $130 for mine.

Note that the Supermicro also requires ECC RAM.

They have open box for SUPERMICRO MBD-X10SLL-F-O (No back panel USB 3.0)
http://www.newegg.com/Product/Product.aspx?Item=N82E16813182819R&

and I already have an extra 4 port SATA controller card, so I can wait for SAS when I really need.

Note that the Supermicro also requires ECC RAM

I already bought Crucial 2x8GB ECC UDIMM.

 

[Muadib]

I have this Supermicro board for my unraid NAS:
http://www.newegg.com/Product/Product.aspx?Item=N82E16813182821

No SAS card needed until you go over 8 drives.

 

[hasu]

Finally I built the file server using Supermicro MBD-X10SLM+-F-O (Newegg: N82E16813182823) and 16GB ECC Unbuffered. I settled for Celeron LGA1150 G3420 for its lower power requirement. Overall idle power with just OS hard drive is around 40-45w. With 8 Toshiba 7200RPM hard drives the average power consumption goes up to about 95-100 W, when in use.

Edit: With 8 Spin drives and 2 SSDs my file server consumes 47w from the wall when all the 8 hard drives are spun down.

 

[FlippedBit]

Just my opinion-

If you're looking for reliability in your data storage system (obviously, since you're using server components) I say skip the add-on SATA controller card. They tend to be cheap. I also read a good argument somewhere that the firmware is written by employees who do not suffer any personal financial loss if there's a bug (= less motivation to get it perfect the first time, especially since it can be updated). At least the on-board controllers have to be solid out of the gate.

It's probably more concerning for I/O controllers than for other kinds of expansion cards, because an error in writing leads to a permanently corrupted file.

 

[hasu]

Just my opinion-

If you're looking for reliability in your data storage system (obviously, since you're using server components) I say skip the add-on SATA controller card. They tend to be cheap. I also read a good argument somewhere that the firmware is written by employees who do not suffer any personal financial loss if there's a bug (= less motivation to get it perfect the first time, especially since it can be updated). At least the on-board controllers have to be solid out of the gate.

It's probably more concerning for I/O controllers than for other kinds of expansion cards, because an error in writing leads to a permanently corrupted file.

What is your point? The chip on an add-on card may be the same as that they use on motherboard as well.

OT: In what industry, government or private, do see employees suffer personal financial losses for the bug they caused? Remember "the bug" caused by a bad decision need not be on a piece of software or hardware.

 

[FlippedBit]

What is your point? The chip on an add-on card may be the same as that they use on motherboard as well.

OT: In what industry, government or private, do see employees suffer personal financial losses for the bug they caused? Remember "the bug" caused by a bad decision need not be on a piece of software or hardware.

A valid point, I just think that a motherboard with a bug is a much costlier mistake than a cheap add on card, and so there should be more concern with choosing reliable components from the start. No?

I don't think the argument I was referencing was making a point specifically about firmware developers. I took the point as firmware/drivers are often the weakest point.

Anyway I was just throwing in $0.02. Feel free to toss it back! ]

 

[hasu]

A valid point, I just think that a motherboard with a bug is a much costlier mistake than a cheap add on card, and so there should be more concern with choosing reliable components from the start. No?

I don't think the argument I was referencing was making a point specifically about firmware developers. I took the point as firmware/drivers are often the weakest point.

Anyway I was just throwing in $0.02. Feel free to toss it back! ]

Hmm... I still don't completely follow your point. If motherboard with a bug can prove to be costlier, then we should get a simple motherboard and get a good add-on card, correct?

 

[dave_the_nerd]

It's FUD. Don't worry about it.

SATA controllers are only made by a couple of companies, and they take it seriously. Embedded/firmware programmers are... weird people. They don't make mistakes because they're not getting paid enough. (That's also not standard compensation practice, like, anywhere.)

Aa far as a sata controller fw bug that eats data, I've never heard of that happening, like, ever.