Shub

Distinguished
Apr 23, 2009
8
0
18,510
Hi again all :hello:

So im having a system built and have been asked these questions, im confused and would like to know what configs you use or would choose?

"1) Multiple Hard Drives - ( Raid Configuration )

2) What Raid Config? - Stripe / Mirror / Raid 5 / Raid 10 / 0+1

3) Hard Drive Partitions?"


Im having these drives:

System drives 1000 GB Samsung Spinpoint F1, 7200 rpm, 32MB Cache

Storage Hard drives 1000 GB Samsung Spinpoint F1, 7200 rpm, 32MB Cache

Solid State Drives 80GB Intel X25-M SSD (Read 250MB/s Write 70MB/s)


I have read the Raid FAQ but im still not sure whats best :??:

Any help/advice would be much appreciated!

Thanks, Shub ;)


 

jrst

Distinguished
May 8, 2009
83
0
18,630
4-8x Intel X25-M's RAID 10 with a top-of-the-line hardware RAID controller would be a good start. Even better, make it 8-16x X25-E's. ;)

Seriously... A little more info on your budget and how you plan on using the system would help.
 

Paperdoc

Polypheme
Ambassador
The starting point has to be, what do you plan to use the system for, and what benefit do you expect from RAID? There are several RAID's of increasing capability and complexity. NONE of them is a substitute for a proper backup procedure, although some of the more advanced ones can help you recover from a hardware malfunction without using your back ups IF things go well. Some claim to speed up your disk access, but the difference may not be very large - depends on your applications.

RAID0 distributes your data over 2 disks, using them alternately to handle sequences of many blocks of data. The alternation process can eliminate a lot of the time "lost" in head seek and disk turning operations in sequential read and write situations, so there's some potential for a speed gain. The total disk system capacity is the sum of the sizes of the two drives, so you get all the storage capacity you paid for. On the other hand, splitting your data between two drives means that the BOTH MUST be working - a failure in either one means you lose ALL your data, so the probability of this disaster is twice as high as using no RAID at all.

RAID1 uses two drives, but the second one is simply a complete mirror copy of the first. So as far as capacity goes you get one drive's space, but pay for two drives. There MAY be some small speed-up of disk operations because the system sometimes can use the first drive available and not have to wait for the second one. The most important benefit is data security. For most simple malfunctions the system can warn you and help recover from the problem because it has at all times two good copies of the data, and it checks whether they agree. BUT any "normal"-looking operation, like deleting a file by mistake or an attack by a virus that writes bad data all over the place will happen to both drives at the same time, so both copies get corrupted this way and the data duplication cannot prevent this type of trouble. So do not rely on RAID1 as a substitute for real backups.

RAID5 uses multiple disks (3 to 5, usually) and spreads your data over all of them. Plus, it calculates some extra data and stores that, too. In cases where some of the data is corrupted, the system can use the extra data it created and stored to completely re-construct the missing data and restore everything to original condition, even while the whole system is still in full use. This is certainly a better system than lower RAID's, but it uses more resources and can use up a little more processor time. For example, a classic 5-drive RAID5 array uses 4 drives for data, and one more for the check data. All 5 should be the same size and model for optimal use. And if you're planning to use its major feature, the ability to re-construct itself after a drive failure, you probably should buy up front a sixth drive to match the other five. So you get 4 drives' space, but pay for 5 or 6.

Higher systems like RAID6 or RAID10 are more complex, use more drives, and provide more disaster recovery features. RAID5 and above usually are used only on busy server systems where constant availability and fast disaster recovery with minimal downtime are really important.

I saw a fine example of why RAID does not substitute for backups. It happened in a well-run professional system using RAID5. They also did automated daily incremental backups and weekly full system backups, both to local tape drives and to a remote backup service. One day one of the drives in the array failed., but the system could continue running at slower speed for a while. There was a delay because the equipment supplier did not have a matching hard drive in stock for replacement and it had to come in by air the next day. When it was installed and the recovery feature activated, they found that a SECOND drive also had failed! So the automated recovery could not be done because they did not have a full 4 out of 5 data sets to use. They had to work without the full set of server files while they replaced the second drive and then restored the entire system from backup tapes while still under full use. It took them about 3 days after the second drive had been replaced. But it all worked, and the data lost was only the small fraction from the early part of the first day of failure.

Bottom line is, for most individual home users there is little value in RAID. Some gamers claim RAID0 gives them a response speed edge on their time-critical games. Some people use RAID1 as a first level of automated backup, provided it is really backed up properly, too. Most modern drives are already so fast, however, that neither provides any meaningful speed-up for most desktop users. If you're spending the money on extra drives, consider two options. One is, just use more drives for more disk space. The other is, use one or more drives (maybe a detachable external drive) as your backup location. Just discipline yourself to make regular backups, then VERIFY that they are good and can be used for recovery. If you're conscientious, you will store the External Drive with its backup data off site.