Homerolled NAS Questions

[Sequences]

My parents are in dire need of a network storage solution and I've been volunteered to make one. There is a spare PC that I can use for this endeavor with more than enough specs to serve as a file server.

I have a few questions which, hopefully, someone can answer:

What kind of use environments are WD-RED drives supposed to be used in? I see it always says NAS with RAID. If I am not going to buy a NAS unit, would keeping my file server on the whole time be the same?

I'm thinking of buying a few drives and installing them as standalone mounts. I distrust RAID as a viable mechanism as a stopgap for failures as it is not a backup solution. Once this project is done, I will be searching for an actual backup solution in the form of a pre-built NAS. But that is for another day.

My parents currently have about 6TB worth of data. Is it better to buy more smaller capacity drives or go with the big boys? Most of the deals I see are for multiples of 3TB drives of the WD-Red flavor.

The current OS on the spare PC is Win7. I plan on simply sharing the folders to the workgroup on the home network. Think this is enough?

Thanks.

 

[smitbret]

I have never owned a WD Red and from reviews and failure data reports I would never pay the extra for one. I am using a combo of Seagate 7200.14s, WD Green, old Samsung SpinPoint and a couple of Toshibas that I harvested from good deals on external HDDs. None have failed in two years.... knock on wood. I leave it on 24/7. The Reds have some firmware updates that are supposed to make them more viable for 24/7 server usage, but homeserver usage is usually not much more demanding than simple desktop or tower use so enterprise drives are often not worth the extra investment, especially if you have stuff backed up.

You are right that RAID is not a backup, but it does make sure the array stays up and that data continues to be accessed even with a dead HDD.

Your current plan is fine. For simple home use Win7 and shared folders will work well. You could consider some drive pooling software to kind of consolidate the storage drives and make them appear as just one big, shared drive (i.e. https://stablebit.com/). If you wanted to consider RAID for uptime purposes, SnapRAID is free and runs on top of Win 7 or a Linux OS. FlexRAID will create a RAID array and take care drive-pooling if you are willing to pay for those features.

 

[Carson Dyle]

What kind of use environments are WD-RED drives supposed to be used in? I see it always says NAS with RAID. If I am not going to buy a NAS unit, would keeping my file server on the whole time be the same?

They can also be used in a non-RAID JBOD application. I'm using several like this now.

I'm thinking of buying a few drives and installing them as standalone mounts. I distrust RAID as a viable mechanism as a stopgap for failures as it is not a backup solution. Once this project is done, I will be searching for an actual backup solution in the form of a pre-built NAS. But that is for another day.

It's not terribly wise to put it off for "another day". You should have a backup solution from day one. No reason why your backup can't be just another PC like you're putting together. Or external drives. Or even the cloud.

My parents currently have about 6TB worth of data. Is it better to buy more smaller capacity drives or go with the big boys? Most of the deals I see are for multiples of 3TB drives of the WD-Red flavor.

In terms of cost per TB, the sweet spot right now is in 3TB drives, with some occasional great deals on 4TB drives. Larger drives are more convenient if you're doing JBOD, as you have fewer storage spaces and it can be easier to find things. Or, you could use some type of storage pooling software to treat all of the drives as one.

The current OS on the spare PC is Win7. I plan on simply sharing the folders to the workgroup on the home network. Think this is enough?

Yes.

 

[Sequences]

I have never owned a WD Red and from reviews and failure data reports I would never pay the extra for one.

There's always something wrong with someone's review/report/writeup. A few of the major buying factors for me are popularity and warranty length. It seems like Red's have 3 years of warranty, which is the minimum I'd like to have. Of course, given the right price, I suspect 2 years would be acceptable as well, provided I backup in a timely fashion.

You are right that RAID is not a backup, but it does make sure the array stays up and that data continues to be accessed even with a dead HDD.

Can you elaborate a little on that?

It's not terribly wise to put it off for "another day". You should have a backup solution from day one. No reason why your backup can't be just another PC like you're putting together. Or external drives. Or even the cloud.

Yeah... I need a kick once in a while. I just want to be sure I put something working in place for my parents. Getting this setup is kind of expensive. The data currently resides in many places. The plan is consolidating into a centralized setup with the appropriate backup plan.

Thanks for the responses so far.

 

[Essence_of_War]

Can you elaborate a little on that?

Your array is at location A. You are at location B streaming some fine media that you stored on A. Suddenly a drive fails! Will your fine media streaming be interrupted?

 

[dave_the_nerd]

My parents are in dire need of a network storage solution and I've been volunteered to make one. There is a spare PC that I can use for this endeavor with more than enough specs to serve as a file server.

I have a few questions which, hopefully, someone can answer:

What kind of use environments are WD-RED drives supposed to be used in? I see it always says NAS with RAID. If I am not going to buy a NAS unit, would keeping my file server on the whole time be the same?

Yes. 24/7 operation is the key thing.

I'm thinking of buying a few drives and installing them as standalone mounts. I distrust RAID as a viable mechanism as a stopgap for failures as it is not a backup solution. Once this project is done, I will be searching for an actual backup solution in the form of a pre-built NAS. But that is for another day.

Six of one, half a dozen of the other. RAID makes certain things easier to deal with (storage management). It is actually a pretty effective "stopgap" for failures (services remain available while RAID is rebuilding). But you're right about it not being a backup - ideally, you'd have a RAID array, and another one to back it up. (A prebuilt NAS is no more of a backup than the second PC you're building up.)

Most of the complaints about RAID only really apply to RAID-5 anyway.

But regardless, yeah, nobody's holding a gun to your head and making you use RAID. Throw a couple of individual HDDs into an old PC and back the whole thing up to Crashplan. Done deal. (I wouldn't just have two mirrored NAS at the same site. If the data is THAT important, than one should be offsite lest a fire or lightning strike kill them both at the same time.)

My parents currently have about 6TB worth of data. Is it better to buy more smaller capacity drives or go with the big boys? Most of the deals I see are for multiples of 3TB drives of the WD-Red flavor.

You will be somewhat restricted by the configuration of the PC you're re-purposing. (If it only has ___ HDD slots or ___ SATA ports) But 3TB drives are currently the best $/GB deal.

http://pcpartpicker.com/parts/internal-hard-drive/#sort=a6

(Internal HDDs sorted by price/GB.)

That "sweet spot" tends to lag behind the latest/greatest HD sizes.

The current OS on the spare PC is Win7. I plan on simply sharing the folders to the workgroup on the home network. Think this is enough?

That is adequate for dealing with just a couple of other Windows clients. If you were in a mixed Windows/Mac/Linux environment, something like Nas4Free or unRAID would be better.

Thanks.

 

[dave_the_nerd]

Can you elaborate a little on that?

RAID stands for "Redundant Array of Inexpensive Disks." (Or "Independent" depending who you ask.) The "Redundant" is a key word.

A text file:

"1234567890"

A hard drive:

[ ]

That text file stored on a single hard drive:

[ "1234567890" ]

That text file stored on a RAID-0 (Striped):

[ "12345" ] [ "67890" ]

If either HD fails, the file is useless because only half the data is there. But it's faster because you can read or write both halves of the file at the same time. The more disks in a RAID-0, the faster it goes.

A four disk strip would like like this:

[ "123" ] [ "456" ] [ "789" ] [ "0xx" ]

With just a little filler at the end or something. And it would be faster-est.

Many people don't consider this truly RAID, since it's an array of independent disks, but there's no redundancy.

That text file stored on a RAID-1 (Mirrored RAID):

[ "1234567890" ] [ "1234567890" ]

If either HD fails the file can still be accessed. After being replaced, the remaining HD would be re-copied to the new HDD.

That text file stored on a RAID-5 (Parity RAID):

[ "12345" ] [ "67890" ] [ "79135" ]

The third drive holds calculated parity information. If one of the drives fails, the missing value can be restoring by adding/XOring, or whatever magic math happens, and the missing data rebuilt. (7-6=1, 9-7=2, 1-8=3, 3-9=4, 5-0=5)

Writing is slower, because you have to do math. But reading data back can be as fast as a simple striped RAID. (Because you can get all the pieces of the file at the same time.)

The basic concept scales well, and can be expanded to RAID-6 (double parity, allows any two drives to fail instead of just the one). The downsides to parity RAID tend to be related to write performance and CPU overhead.

Traditional RAID-5 (not such a big problem with similar schemes like RAID-Z or some software RAIDs) is also prone to eating itself if there's a read error during a rebuild - the bigger the disks, the more likely there will be a read error. And there's that "write hole" thing.

Realistically, you shouldn't be deploying RAID-5 in a production environment, because it's been replaced by better and more awesomer things, but the basic principles and advantages (N disks, capacity of N-1, fast striped reads) still apply.

What this all means is that on a Mirrored-RAID or Parity-RAID array, the data is still accessible after a single drive failure. Which can be helpful. This is normally referred to as "degraded" operation, since the array is no longer "Redundant."

 

[Data-Medics]

Wow, Dave, you have a lot more patience than I have to type all that out. I'd just pull one of these:

Read here to learn what RAID is: http://www.data-medics.com/choosing-which-type-of-raid-to-use.html

 

[Carson Dyle]

OP: Your initial instincts are spot on.

I distrust RAID as a viable mechanism as a stopgap for failures as it is not a backup solution.

The number one purpose of RAID is to maximize uptime in mission critical situations. For the vast majority of home server applications, uptime is far down the list of priorities. Data security should be foremost. Keep good backups. Period. You need them with or without RAID.

 

[Sequences]

Thank you all for the patient explanations of the R in RAID. I don't know why I had such a brain-fart to smitbret's original comment, I should have known what he was talking about.

One of my biggest fears with RAID is whether or not my data is safe if the controller fails. Can someone comment on this?

Uptime and performance are not priority in this case, data integrity and capacity are.

 

[dave_the_nerd]

Thank you all for the patient explanations of the R in RAID. I don't know why I had such a brain-fart to smitbret's original comment, I should have known what he was talking about.

One of my biggest fears with RAID is whether or not my data is safe if the controller fails. Can someone comment on this?

Sure. If you're using a hardware raid controller, just replace it with a similar model and import/recover the array.

Although recovering from backups may be easier, depending.

But any more it's really just software RAID. Those are portable between machines and more or less controller-independent.

Uptime and performance are not priority in this case, data integrity and capacity are.

 

[dave_the_nerd]

Wow, Dave, you have a lot more patience than I have to type all that out. I'd just pull one of these:

Read here to learn what RAID is: http://www.data-medics.com/choosing-which-type-of-raid-to-use.html

I have to - my company knowledge base is internal.

 

[tehach]

Recommend NAS4Free installed on a discrete drive. Solid platform and as mentioned upthread handles mixed OS well.