nas CPU and RAM for NAS?

[BobyTheDragon]

So small question about a NAS I am thinking of building (After losing a action for a 8 core 16 thread server... ;-; ).

 

Would a Pentium (G4500 or G4400) and 4gb's of ram be able to look after 2 4TB WD Red HDD in raid 1?

 

[sunil6512]

My NAS is running 8TB raid 0 on an Intel atom CPU and 4GB ram just fine. If you want to run a Plex server on it then get a better CPU.

[BobyTheDragon]

Just now, sunil6512 said:

My NAS is running 8TB raid 0 on an Intel atom CPU and 4GB ram just fine. If you want to run a Plex server on it then get a better CPU.

Don't know what a Plex Server is

I was just gona get unraid and pray to god I get it right first time.

[sunil6512]

1 minute ago, Marshall212 said:

Don't know what a Plex Server is

I was just gona get unraid and pray to god I get it right first time.

uRraid is quite expensive , why not try Free Nas ? 

[BobyTheDragon]

Just now, sunil6512 said:

uRraid is quite expensive , why not try Free Nas ? 

Oh i'm an idiot I thought Free Nas was a nickname for unRaid LOL

[SerialAceTU1]

Generally NASes require less powerful processors and less RAM than a consumer-oriented computer.

Also depends on how many HDDs you have in the NAS

 

 

 

 

[BobyTheDragon]

Just now, SerialAceTU1 said:

Generally NASes require less powerful processors and less RAM than a consumer-oriented computer.

Also depends on how many HDDs you have in the NAS

 

 

 

 

I will be adding more hard drives after a while. Was thinking about 100 Terabytes or so... Na jk might go to like 16TB (4x4 HDDS in raid 1)

[Jarsky]

13 hours ago, Marshall212 said:

I will be adding more hard drives after a while. Was thinking about 100 Terabytes or so... Na jk might go to like 16TB (4x4 HDDS in raid 1)

RAID1 is mirror. If you have 4 x 4TB drives you will have 4TB (3.6TB formatted) space with 3 hot-spare drives - meaning you can lose up to 3 drives

If you run a RAID1+0 then you'll have a mirrored stripe giving you 8TB (7.2TB formatted) space - and will give you 1 hot-spare drive on each stripe meaning you can lose 2 of the *right* drives

The most popular for 4 disks would be a RAID5, giving you 12TB (10.9TB formatted) space, with 1 parity drive - meaning you can lose 1 drive, replace it and the array will rebuild from parity across the remaining drives.

[beavo451]

7 hours ago, Jarsky said:

RAID1 is mirror. If you have 4 x 4TB drives you will have 4TB (3.6TB formatted) space with 3 hot-spare drives - meaning you can lose up to 3 drives

If you run a RAID1+0 then you'll have a mirrored stripe giving you 8TB (7.2TB formatted) space - and will give you 1 hot-spare drive on each stripe meaning you can lose 2 of the *right* drives

The most popular for 4 disks would be a RAID5, giving you 12TB (10.9TB formatted) space, with 1 parity drive - meaning you can lose 1 drive, replace it and the array will rebuild from parity across the remaining drives.

While you are generally correct, you are using the term "hot-spare" incorrectly. 

 

A hot spare drive or hot standby drive is powered and connected to the array, but does not hold any data. If the controller detects a drive failure, it immediately switches from the failed drive to the hot spare and begins an array rebuild.

[Jarsky]

Well yeah, I was using it in a term that makes sense for a fellow kiwi 

[beavo451]

9 hours ago, Jarsky said:

Well yeah, I was using it in a term that makes sense for a fellow kiwi 

How about explaining it correctly then? Using incorrect terminology causes confusion.

[BobyTheDragon]

15 hours ago, Jarsky said:

*snip*

 

6 hours ago, beavo451 said:

*snip*

Yea I'm not expert on raids so you kinda lost me.

Lets see what Google says because what you said just bounced off me like a tennis ball...

 

EDIT: (Talking about Raid 5) So what I see is that the drives have a partial copy each other? like raid 0? When one drive fails the other 3 drives can rebuild the information to a new drive but its also acts like raid 1 because all 4 drives can have the same info. Allowing Read and Write to be quick and simple... Yea if I got that wrong there holy crap I need to do more research on RAID...

[beavo451]

1 hour ago, Marshall212 said:

 

Yea I'm not expert on raids so you kinda lost me.

Lets see what Google says because what you said just bounced off me like a tennis ball...

 

EDIT: (Talking about Raid 5) So what I see is that the drives have a partial copy each other? like raid 0? When one drive fails the other 3 drives can rebuild the information to a new drive but its also acts like raid 1 because all 4 drives can have the same info. Allowing Read and Write to be quick and simple... Yea if I got that wrong there holy crap I need to do more research on RAID...

Ok, I will make this very basic and simplified with analogies.

 

Firstly, how data is stored:

 

Normally, data is stored in "blocks". So, a file is made up of bits (and subsequently bytes) of data. This data is grouped into blocks to be stored on media (like a hard drive).

 

So, think of data like sand. You build a sand castle and you want to store that sand castle (saving a file). You take that sand castle (a file), break it down and put it in a bunch of little, tupperware containers (blocks), and you put the containers in your closet (hard drive). If you want to look/play/work (opening the file) with the sand castle (file), you take the containers (blocks) out of the closet (hard drive), and reassemble the sand into a sand castle.

 

Secondly, what is the point of RAID?

 

RAID protects against drive failure so that you don't loose all your information (except for RAID 0). So, say your closet (hard drive) catches on fire and burns down. All the stuff or information in it is destroyed.

 

Thirdly, how does RAID work?

 

So using our sand and sand castle analogy...

 

RAID 0 - Stripes data across all member disks. DOES NOT PROTECT AGAINST DRIVE FAILURE! RAID 0 is normally done to increase performance.

So say you have two closets (two hard drives). You could fill one closet up with stuff until it is full and then use your second closet. Or you can split each container of sand in 2 and put half of a container in each closet at the same time. So, if one closet goes up in flames, you only have a bunch of half containers.

 

RAID 1 - Mirrors data. You have to have an even number of disks for RAID 1. Can withstand (total number of drives - 1) disk failure(s).

Take those same two closets. You put your containers of sand into one closet. At the same time, you magically create an exact copy of each container and put it into the second closet. So, if one closet goes up in flames, you still have one closet of stuff. When you replace the burned up closet (failed hard drive), you automatically start copying each container and putting it into the new closet (rebuild). So, if you have 5 closets, those 5 closets will have exactly the same contents. And because they are all exactly the same, 4 of the closets can go up in flames without data loss.

 

RAID 5 - Data striping + parity block. Requires at least 3 drives. Can withstand 1 drive failure regardless of the number of drives.

So, now you have three closets. Think of the RAID 0 above with two disks. The containers are stored in two closets just like the two disk RAID 0 example above. Now you look at each container of sand and create a sort of blueprint if your will (partiy block. Note: this is not what exactly happens, but for the sake of understanding and the analogy it works). You store this blueprint in the third closet. For each container of sand, you create a blueprint for that container. The blueprint is stored in the third closet. But, for every data block, the location of the blueprint is rotated to another closet. So when one closet goes up in flames, you can calculate what data is missing and recreate the missing data when you replace the failed closet.

 

RAID 6 - Data striping + parity block + another parity block. Requires at least 4 drives. Can withstand 2 drive failures regardless of the number of drives.

Same as RAID 5 as above, but now you are making 2 "blueprints" that are stored on separate drives.

 

Hot Spares:

 

Okay, so your data is protected against drive failure. But, what happens when a drive actually fails? The array rebuilds itself when the failed drive is replaced by the operator. Of course, this takes time. RAID 1 has to recopy the data. RAID 5 and RAID 6 has to recalculate the parity blocks (redraw the blueprints). How much time this takes depends on the size of the drives and the amount of data. During this rebuild time, the performance of the array is degraded and is vulnerable to another disk failure. Say you have a disk failure in a RAID 5. You replace the failed drive and the RAID starts a rebuild. If another drive fails during the rebuild, data is lost because there is not enough information to recreate what was stored. The longer an array is in this degraded/rebuilding stage, the more vulnerable to another disk failure and data loss it becomes.

 

A disk failure can happen at anytime. So if you are running a business and a disk fails on a Friday evening right after everybody leaves, the disk failure won't be discovered until Monday morning if you don't have anybody monitoring it. A hot spare disk is a disk that is connected to the system, but is not an active drive. So when the RAID controller detects a failed disk, it switches to the hot spare disk (as if a person physically replaced the failed disk immediately) and rebuilds the array. Instead of a vulnerability time of the weekend + rebuild time, you just have the rebuild time.

[Rob_LC]

I have a small unraid server with a i3-550 and 4GB ram (with 6 SATA ports), so your spec would be fine.

If you just use it for raid, it will run on 2GB ram easily. I run a docker with zoneminder recording my security cam, a print redirector to my usb printer and sometimes a Server 2016 test VM in 2GB of ram on top of that, because unRaid can host VM's.

 

For storage I use 2x 3TB drives, giving 3TB storage. If I add one more 3TB, i would have 6TB storage. unRaid uses 1 parity drive, and if any 1 drive fails you don't loose data. The more drives you use, the higher the percentage of the storage is used for storage. Its very suitable for home use and only $59US (trial free for a month).

Even better, if you have 4x 4TB disks, you have 1x parity, 3x data which is 12TB of storage. If you were unlucky enough that both the parity and one data disk failed at the same time, the two remaining disks will have valid files remaining - unRaid does not stripe or split the files across disks to improve performance.

 

 

Edited April 5, 2016 by Rob_LC