Hard Drives / Raid / etc

[Spit-wad]

So sometime in the near future I will be coming across a lot of data. More than my 500GB Seagate can hold. I want to have enough room for all of this data, and it will be extremely important for me to back it up.

I'm still quite the "newb" at most hardware topics... so what are my options for doing this?

I don't mind storing all of this data in my main PC, but I could also set up a separate home server... don't really know what the benefits of each would be.

Anyways, I read up on hard drives and raid. I was thinking a Raid 5 setup might be my best option? My estimate is that I'll need about 2TB... so to do that, I would need 3 1TB HDD's, correct?

Any other ways to do this that are worth looking into?

Halp!
Thanks

 

[Lexluethar]

I believe a RAID 5 setup would be your best option. You would need a RAID controller and three identical drives. I would suggest doing three 1 TB drives, but you could do two 500 GB drives. With RAID 5 essentially you get the use of two drives worth of space (not actually on just two drives). RAID 5 uses parity to backup the information, one drives worth of data is the parity information, then the other two you get to use for information.

This would be my suggestion.

 

[Spit-wad]

What does the RAID controller do? Recommend any particular one?
Is there a guide somewhere on how to setup RAID?
How do I recover data if one drive fails?
(Doesn't RAID 5 require 3 or more drives? If I only had 2, I would do RAID 1?)

Thanks

 

[Lexluethar]

A RAID controller controls the information going from HD to HD and to your system. I would recommend using a controller instead of onboard (new motherboards come with onboard RAID 0/1) because RAID does require some computing power. So instead of eating up CPU clock cycles with RAID functions, you can just purchase a RAID controller to do all the thinking for you, and your system isn't bogged down at all.

Take a look at newegg and read some reviews.

When you purchase the controller it should come with a guide and software.

If one drive fails you replace the drive w/ a new one (identical in size) and with the software you 'rebuild' the RAID array. Populating the new drive w/ some parity information and data.

RAID 5 requires 3 drives. Reason being is you split the information amongs the drives including parity information. As we all know the data is mathmatical, so with RAID you can have the equation 1+1=2. 1 being some of the information, the other 1 being the other half, and 2 being the 'parity' of both bits of information. So say you lose a drive and now all you have is 1+?=2, well you know the difference of 1 and 2 is 1, so you can replace the data. This is what the RAID controller does and the software that comes with it.

If you only have 2 drives you can do RAID 0 or 1. RAID 0 gives you no data security and uses disk striping allowing the system to access the data faster. RAID 1 is mirroring, where one disk is an identical copy of the other.

 

[mikesgroovin]

I would also suggest RAID5. You can rebuild the data fairly easily but it takes time to rebuild (a lot more time than rebuilding a mirror) If you are using mid-range performing drives, perhaps 7200.10 drives... you can expect a 200GB rebuilding process to take a GOOD hour or so. And thats only 200GB)

As far as where to store it, thats really up to you. You can setup something similar to a RAID5 in it's own little chassis (similar to a NAS) or use a computer for it with an internal controller. Really just depends on certain facts:

1) Will the data change over time
2) Heat
3) Other drive uses
4) Mobility
5) Security
6) Other uses for the computer housing/holding the data

In my experiences with RAID5 arrays, we would typically use a 5-bay available chassis and use 4 drives in the array and have 1 available as a hot spare. This was a while ago when 1tb drives were unheard of and RAID arrays in home systems were NEVER common. So, the budget typically had the availability for good, reliable hardware. I would slap 5 drives in the SCSI chassis, tell it which was the SCSI0,1,2,3 and spare and then go away. It just took care of itself without lifting a finger unless every once in a blue moon, there would be a bad drive in the log. Then just pinch out SCSI3 and slap in a new drive. But yea, if you ever need to watch or wait for your array to rebuild in order to add to it... it's going to take a LONG time for 2TB worth of data.

 

[Spit-wad]

1)The data will change over time... new data will be added very often
2)I have an Antec 900 and so far I have not experienced any heat issues with any of my components.
3)I have a 500GB currently, which I would like to continue using as I currently am, to hold my personal stuff. I just need 1.5-2TB of storage space which I will keep separate - this will be data that others will need to access.
4)Mobility is not a huge concern. I have to move twice a year and I maybe take it to one or two LAN parties per year... usually it's sitting in the same spot.
5)??? Explain plz???
6)It's my primary computer. I game on it, watch movies, listen to music, etc. I'll keep using my 500GB hdd for my personal uses, and the other ~2TB will be for my home business, which should be kept completely separate.

I think I would actually like it to have it's own chassis and use it as a little server (as long as it wasn't the size of a mid case), but I really need to keep costs down on this project. What's the best way to do this?

I was thinking about getting 3 of these:
Newegg.com - SAMSUNG Spinpoint F1 HD103UJ 1TB 7200 RPM 32MB Cache SATA 3.0Gb/s Hard Drive - Internal Hard Drives
~$405, 2TB available space

Or 2 of these and doing RAID 1:
Newegg.com - Seagate Barracuda 7200.11 ST31500341AS 1.5TB 7200 RPM 32MB Cache SATA 3.0Gb/s Hard Drive (bare drive) - Internal Hard Drives
~$380, 1.5TB available space

I would probably just do RAID 5, seems like more bang for the buck.

Anyways, what would be involved in me putting this in it's own chassis? I will need to set it up as a direct connect hub for other people on my LAN to access. Would I just have to build a whole computer?

Obviously we'll hope that I never have to rebuild the array, but is there really any way around this? I wouldn't mind waiting like 20 hours as long as it works.