I am going to be building a computer system for a client which will contain alot of information and bits of software running 24/7.
I have decided that I should get them a RAID system. will be using WD HDD's (1TB drive)
the reason for RAID is that I believe, if a HDD goes down, at least we got the other as a backup and still can run off it, or rebuild the system?
can someone enlighten me on what exactly are the benefits of RAID, which is better and how to configure a RAID system and how many HDD's are required? What about when a disk goes down, how would I recover the system?
thanks. I still need clarification than just reading a wiki article as I have done before, which is why I posted the question on a forum in the first place......
which is recommended for performance and redundancy?
how many HDD's are required for that RAID configuration?
in a recovery/rebuild - how does one rebuild the RAID?
Rebuilding the array will really depend on the controller. Normally replace the failed drive and use the controller interface at boot to rebuild the array.
Raid0 - Stripe set NO REDUNDANCY - some perfomorance increase. Requires two or more disks. One drive fails, all data is lost. Some performance increase
Raid 1 Mirroring - All data is replicated to the second disk. Disks appear to OS as only one logical drive. Two disks. 2 x 1 TB disks = 1 TB logical disk. No performance increase
Raid 5 Strip with Parity - Requires three or more disks. One disk can fail and the others have parity information that allows them to rebuild the failed disk. Slight write speed decrease, slight read increase 3 x 1 TB disks = 2 TB useable disk space.
All drives in any raid must be the same capacity, or the raid will use the capacity of the smallest disk.
so in future if more storage is required to add into the existing system, what needs to be done?
if RAID1 is chosen, and say if a disk fails - what happens? Would the system still function as normal? if not - after replacing the failed drive with another working drive means that the drive must be rebuilt from the controller interface and everything would be back to normal?
On Raid 1 and 5 both, you will get a warning the drive has failed, but the system should continue to run. Not sure about some of the cheap consumer controllers built onto MoBos today, but every controller I have used has continued to function - of course they were SCSI based controllers in servers, I'm not a big fan of RAID on a home system - backups are a better solution. RAID will only protect against physical drive failure, it is not a replacement for backups! If the file system gets corrupt the data will require recovery like a single drive, but is a lot tougher to do, and a lot more costly if you have to go to pro recovery. In most cases, a good backup solution will provide as much protection with less hassle and just use single drives. Other than a server, where uptime is critical I don't really see much point in running it. We see a lot of raids come through for recovery some of them cost over $5000 for even a 3 or 4 disk array.
The advantage of RAID is that the system continues in operation, even after a disk failure. The system is only down during the disk replacement. With hot-swap drives (available on high end server hardware) there is no downtime even during disk replacement. For systems that must maintain continuous service this is an enormous advantage.
Of course you must still maintain regular backups. RAID protects you from data loss due to disk failure only.
Raid one is mirroring, a complete second set of data on both drives, Raid 5 uses striping, data is written in stripes across all drives in teh set and a parity value is calculated and stored also across the drives in the stripes. The parity value is calculated by x-oring the previous stripes, normally one less that the number of disks. Since the parity is distributed, and the last parity strip is calcultaed into the next, any one disk can fail, and the others can recreate that data based on the x-ored data in the parity strip. The performance is a bit slower since the math must be performed to calculate the parity value as the data is written. high end RAID cards will have a dedicated processor to handle that calculation, lower end add-on cards, and as far as I know all MoBo enabled raid use the system CPU and memory to do that calculation so there is some slowdown there, as well as writing the extra data for parity. There is no perfect answer for which one is best, they have different applications, and what the system is being used for is one of the considerations.
Nope Raid 1 is a complete mirror copy of all data on each drive, raid 5 stripes the data across the disks in chunks of usually 64k or 128k stripes, with a calculated parity strip so the data in the previous stripes can be recreated, but it r3equires the calculation to be done to recreate it.
The practical purpose of all RAID systems (except RAID 0) is the same - the system will continue to operate in the event of a single drive failure. Each RAID type has differing hardware requirements, performance characteristics, and methods of recovery.
This should be possible with RAID 1. It will not be possible with other forms. RAID levels 2-4 are rarely used and are most controllers do not support them.
Be sure to have a backup of all important data before beginning and follow the configuration instructions. These will vary widely depending on the controller in use. With any type of RAID be sure you understand the potential problems (particularly with RAID 5) and how to do a recovery in the event of failure. These will vary depending on the controller.
A forum community dedicated to tech experts and enthusiasts. Come join the discussion about articles, computer security, Mac, Microsoft, Linux, hardware, networking, gaming, reviews, accessories, and more!