Would anyone have any thoughts on this hardware?

 

The MB isn't listed specifically on the MB list but seems to have good reviews on NewEgg and has 8 SATA ports, two regular PCI-E, etc. And sounds like the processor should be enough for unRaid.

 

Case and Icy Dock are same as the RB-1200. For the RAM, I saw someone mention that having one at 2GB would be ok since it'd let me upgrade to 4 later if needed and the dual channel won't help that much?

 

For a PSU, I do have a spare Corsair 750 from a project that never happened, or I could buy a 450 since the 750 is such overkill.

 

Any thoughts? I'm thinking the 3TB of actual storage will last me a bit and can throw in some older 500GB drives I have later on if I need more. Or just get some more 1.5TB ones. And this way I can start with the free unRaid before upgrading..

 

Thanks,

Shawn

 

link is broken for me so can't see the motherboard.

 

2gb ram is fine and 450w PSU should also be fine unless you are planning to have lots of disks in the future.

Would anyone have any thoughts on this hardware?

 

Yeah, that wish list isn't working for me either.  Please repost direct links to your hardware.

 

For the RAM, I saw someone mention that having one at 2GB would be ok since it'd let me upgrade to 4 later if needed and the dual channel won't help that much?

That is correct, and I believe I'm the one who originally said that.

 

For a PSU, I do have a spare Corsair 750 from a project that never happened, or I could buy a 450 since the 750 is such overkill.

Up to you.  750 W is definitely overkill, but if you already have it, you may as well use it.  Saves you a bit of investment right now, and gives you room to grow to a monstrous 10+ disk array in the future.

Whoops! Sorry about that..

 

MB is GIGABYTE GA-770TA-UD3.

 

The rest was a COOLER MASTER Centurion 590 case, single Icy Dock 4 in 3, AMD Sempron 140 processor, Kingston 2GB RAM and some Samsung EcoGreen F2 1.5TB drives.

 

Thanks,

Shawn

 

Wow, helluva motherboard for an unRAID build.  Keep in mind that while you may be able to get the SATA 6.0 Gb/s ports working with unRAID, you will not enjoy their performance benefits since the network is far more of a bottleneck.

 

You can certainly save some money by choosing a simpler board, but that's up to you.  The Sempron 140 is a great choice (I'm using one as well), and everything else looks good as well.  Go for it!

...you will not enjoy their performance benefits since the network is far more of a bottleneck.

 

Huh?  How exactly is the network the bottleneck?

 

Copy ram disk to ram disk over the network: ~82MB/s

Copy to unRAID disk over the network: ~23MB/s

 

What are you talking about?

 

Double huh?  Was I wrong about that?  I was under the impression that:

 

GigE (1 Gb/s) is slower than SATA II (3 Gb/s) is slower than SATA III (6 GB/s).

 

Enlighten me?

Enlighten me?

Well, answer this:

1) What is your write speed when you copy stuff over the network to unRAID disk share?

2) What is your write speed when you copy stuff over the network to unRAID RAM disk?

And after you answer 1 and 2, then answer this:

3) How can you then say that the network is the bottleneck?

 

I have no idea how to set up a RAM disk, nor do I really care to since it won't affect my day-to-day unRAID use at all.

 

So you are saying that a file transfer from a PC over a GigE network to a SATA III drive in an unRAID server will be faster than the same file transfer to a SATA II drive?  I guess I'll take your word for it, but I really don't understand how that is possible.

 

This is why I expect SSDs won't be any faster in the unRAID environment either.

 

If I'm wrong, please explain it to me.

So you are saying that a file transfer from a PC over a GigE network to a SATA III drive in an unRAID server will be faster than the same file transfer to a SATA II drive?

I never said anything of that sort! That's a different topic, and I'm not going there now.

What I said was, I objected to you stating that on unRAID the network is the bottleneck, when in fact the disks are the bottleneck.

Wow, now I'm really confused.  If the disks are the bottleneck, then SSDs or SATA III disks should be faster in unRAID than SATA II disks, right?

Double huh?  Was I wrong about that?  I was under the impression that:

 

GigE (1 Gb/s) is slower than SATA II (3 Gb/s) is slower than SATA III (6 GB/s).

 

Enlighten me?

 

Just because SATA II has bandwidth of 3 Gb/s does not mean the disks are capable of that.

I wonder if the ASUS M4A78L-M would be more sensible. It has less SATA (6), less PCI and no firewire or eSata but it is cheaper and has on-board video (so I don't have to buy a video card and won't take up a pci slot).

 

I wonder if the ASUS M4A78L-M would be more sensible. It has less SATA (6), less PCI and no firewire or eSata but it is cheaper and has on-board video (so I don't have to buy a video card and won't take up a pci slot).

 

 

I had also looked at that one recently.  It looks like a good board and the chipset and lan controller seem to show up in different b oards that are on the hardware list.  The previous board with 8 sata ports looks ok if you think you will want that.  For that one just buy a cheapo PCI video card that is supported because the video only displays the console prompt.

Bear in mind I'm pretty new to unRAID so rely on more seasoned if it's offered.

Enlighten me?

Well, answer this:

1) What is your write speed when you copy stuff over the network to unRAID disk share?

2) What is your write speed when you copy stuff over the network to unRAID RAM disk?

And after you answer 1 and 2, then answer this:

3) How can you then say that the network is the bottleneck?

 

The reason that it is slower copying to the unRAID disk share vs. a RAM disk is because of the overhead required to calculate and write parity.  If you have a really fast parity drive, you will likely run into the limits of your network.  Also, if you need better write performance, you should look into having a cache drive.  A cache drive is a buffer: when you copy something to your unRAID server, it will write to the cache drive now, then copy over to your parity-protected array later.  See http://lime-technology.com/wiki/index.php?title=Improving_unRAID_Performance for more information.

 

Unless you are planning to use SSDs for your array, SATA 3.0 Gigabit/second is way faster than any current rotational media (aka hard drives).  Let me explain:

 

SATA 3.0 Gigabits per second = 3000 Megabits per second = 375 Megabytes per second (bits versus bytes, divide bits by 8 to get bytes).  If you pop over to a review site for current hard drives, you will find that very few go over even 150 Megabytes per second.

The reason that it is slower copying to the unRAID disk share vs. a RAM disk is because of the overhead required to calculate and write parity.

Do you serously think that I don't know that?  You missed the point.

I was simply saying that between the disks and the network, it's not the network that's the bottleneck. The disks are.

 

If you have a really fast parity drive, you will likely run into the limits of your network.

Now that is simply not true.  Sorry.  Not with the current unRAID, and now with the current hardware.

 

The reason that it is slower copying to the unRAID disk share vs. a RAM disk is because of the overhead required to calculate and write parity.

Do you serously think that I don't know that?  You missed the point.

I was simply saying that between the disks and the network, it's not the network that's the bottleneck. The disks are.

What I was getting at is that the way unRAID works is the bottleneck - if you wanted faster performance, you should set up a RAID-5 (or other RAID type) in hardware.  If you set up a cache drive, you'll get rid of a lot of the performance difference.

 

Anyway, the core point of my post was to say that SATA 6.0 Gb/s is not worth worrying about with the current speeds of hard drives.  Nearly all current (magnetic, rotational, non-SSD) hard drives do not even max out SATA 3.0 Gb/s.  You could easily get very similar performance with a less expensive motherboard.

 

If you have a really fast parity drive, you will likely run into the limits of your network.

Now that is simply not true.  Sorry.  Not with the current unRAID, and now with the current hardware.

Admittedly I've never personally tried (or cared about) a really fast parity drive, so I could be wrong.  (edit: I am wrong.)  I remember reading some setups that seemed to get close to maxing the network performance, but I could've read wrong (or they could've had a cache drive).  (edit: I read wrong.  They had a cache drive, none of the ones without cache drives got anywhere close to maxing network performance)

The reason that it is slower copying to the unRAID disk share vs. a RAM disk is because of the overhead required to calculate and write parity.

Do you serously think that I don't know that?  You missed the point.

I was simply saying that between the disks and the network, it's not the network that's the bottleneck. The disks are.

What I was getting at is that the way unRAID works is the bottleneck -

BUT THAT WAS EXACTLY MY POINT!  The network is NOT the bottleneck.

(And BTW, if you haven't noticed, I never commented on any SATA6 stuff: that is a separate topic.)

 

Everybody keeps on harping on the desire for a really fast parity drive.  

It DOES NOT help when writing to the array UNLESS the data drive being written is EQUALLY as fast.  

 

Writing to the array can only go as fast as the slowest drive involved.  Since the platter on both the data drive and the parity drive must rotate a full revolution between the read and subsequent write of any given sector, the rotational speed of the disk dictates how fast you can write to the array... SATA-1,2,3,5, 3000 Mb/s... whatever... will have no effect on the time it takes the platter to revolve.

 

The only time a faster parity drive "might" help if two or more slower data drives were being written at the same exact time, but then it might not be able to keep up with seeking back and forth anyway between the involved sectors.

 

Joe L.

 

 

The reason that it is slower copying to the unRAID disk share vs. a RAM disk is because of the overhead required to calculate and write parity.

Do you serously think that I don't know that?  You missed the point.

I was simply saying that between the disks and the network, it's not the network that's the bottleneck. The disks are.

What I was getting at is that the way unRAID works is the bottleneck -

BUT THAT WAS EXACTLY MY POINT!  The network is NOT the bottleneck.

(And BTW, if you haven't noticed, I never commented on any SATA6 stuff: that is a separate topic.)

If you are using one traditional hard drive connected to the network, you will get close to (or reach depending on the drive and the network hardware) the limit of the network.  If you had an SSD (or something that would take advantage of SATA 3.0 fully), you would likely not see much more performance copying files over the network than with one traditional hard drive.  The network would be the bottleneck (it would be fairly close anyway) for one single hard drive.  Using unRAID will net you slower write performance (unless you use a cache drive).  *That* is the point I was trying to make.

 

In an effort to try to get this thread back to its original topic, SATA6 isn't worth paying a premium for if you are going to use traditional hard drives.  You won't be able to enjoy SATA6 performance for a while (unless you use SSDs).  So, the motherboard originally mentioned is probably overkill for the purposes of unRAID - although the OP may want it for other reasons (overclocking because he's doing encoding, lots of RAM supported because he's running VMware, wants to be future proof, etc).

O.M.G.

 

Ok, forget my argument.  I did some more research (and reread the previous posts a little closer).  I'm going to stop trying to patch it and make it seem like I was right.

 

So I was wrong.  I'll gladly admit that.  I'm still really new to ... everything here really.

I have no idea how to set up a RAM disk, nor do I really care to since it won't affect my day-to-day unRAID use at all.

If you have no idea then why say all such things? If you copy on network 20MB/s on unraid disk, and you copy on network 80MB/s on no disk, then how can you say network is botleneck? That is no idea.

 

 

I said that because a RAM disk does not have any relevance in normal unRAID usage.  I'm only interested in real-world numbers, and answering questions like whether a SSD or SATA III drive in an unRAID server will show any performance benefits.  So far that hasn't been demonstrated.

 

However, I understand Purko's point that a RAM disk can be used as proof that the network is not the bottleneck.  Since I haven't tried it myself, I'll just have to take his word on the matter.