Tek-Tips is the largest IT community on the Internet today!
Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!
-
Congratulations John Tel on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!
Quick RAID Question
- Thread starter JeffITman
- Start date
LWComputingMVP
IS-IT--Management
The only RAID option that will allow that is RAID 0 - but that's really foolish in my opinion. RAID is NOT backup - RAID is for redundancy. If one disk fails everything is still running and users are able to continue work without losing information. It's good that you backup, but RAID 0 alone on a server is NOT advisable under most circumstances. (Frankly, I can't think of one instance where I'd use it).
You have to weigh up whether it is more cost effective to have a RAID solution in place or whether your company can afford to have people sitting around while you rebuild a server due to a disk failure.
Since the price of disks are next to nothing in comparision to the loss your business may suffer, I would advise you to purchase another disk and have a 5 disk RAID5 (meaning you would get around the 1Tb you require). Doing this would mean that if a disk was to fail, you would still have a workable server working in interim mode instead of a dead server that needs a rebuild.
--------------------------------------
"Insert funny comment in here!"
--------------------------------------
Since the price of disks are next to nothing in comparision to the loss your business may suffer, I would advise you to purchase another disk and have a 5 disk RAID5 (meaning you would get around the 1Tb you require). Doing this would mean that if a disk was to fail, you would still have a workable server working in interim mode instead of a dead server that needs a rebuild.
--------------------------------------
"Insert funny comment in here!"
--------------------------------------
I totally agree with the above advice. Go with RAID 5.
OR
Stripe 2 drives and then mirror them to the other two if your RAID card will support that.
I hope you find this post helpful.
Regards,
Mark
Check out my scripting solutions at
OR
Stripe 2 drives and then mirror them to the other two if your RAID card will support that.
I hope you find this post helpful.
Regards,
Mark
Check out my scripting solutions at
All good points.
The technical answer to the question is RAID 0, but it's a really bad idea. If a drive crashes you will lose everything that was written since the last backup, and the server will be unavailable to users as well. Say you're using 250 GB drives, you will need only 4 of them.
RAID 5 is a better answer, because it is the cheapest way to get a redundant array of a given size (unless you can do a mirror with only 2 disks). However, RAID 5 isn't suitable for all situations because of it's relatively poor write performance (all forms of parity RAID suffer from this). If you're using 250 GB drives, you will need 5 of them plus a RAID controller that does RAID 5 (most servers have this already).
RAID 10 offers much more write performance, but at twice the cost. But it is also much more fault tolerant. If you are using 250 GB drives you would need 8 drives to build this array. But you could potentially suffer 4 drive failures and still have a working array.
Normally when I size an array I have to look at how it's going to be used. Will it be used for an application that is I/O intensive (databases and email for example), or will the application just need lots of space (file server)? Will the I/O operations be largely reads, writes, or an even mix of both? Assuming the storage controller provides enough bandwidth to the disks, how many disks will you need to reach the required performance levels? Disk space is cheap, but to get more performance you have to add more spindles (disks) and stripe the data across them. All of that needs to be factored into the equation.
The technical answer to the question is RAID 0, but it's a really bad idea. If a drive crashes you will lose everything that was written since the last backup, and the server will be unavailable to users as well. Say you're using 250 GB drives, you will need only 4 of them.
RAID 5 is a better answer, because it is the cheapest way to get a redundant array of a given size (unless you can do a mirror with only 2 disks). However, RAID 5 isn't suitable for all situations because of it's relatively poor write performance (all forms of parity RAID suffer from this). If you're using 250 GB drives, you will need 5 of them plus a RAID controller that does RAID 5 (most servers have this already).
RAID 10 offers much more write performance, but at twice the cost. But it is also much more fault tolerant. If you are using 250 GB drives you would need 8 drives to build this array. But you could potentially suffer 4 drive failures and still have a working array.
Normally when I size an array I have to look at how it's going to be used. Will it be used for an application that is I/O intensive (databases and email for example), or will the application just need lots of space (file server)? Will the I/O operations be largely reads, writes, or an even mix of both? Assuming the storage controller provides enough bandwidth to the disks, how many disks will you need to reach the required performance levels? Disk space is cheap, but to get more performance you have to add more spindles (disks) and stripe the data across them. All of that needs to be factored into the equation.
- Status
Similar threads
- Locked
- Question