Discussion Forums > Technology

Partitioning Hard drives: Healthy or not?

<< < (3/4) > >>

fubuu11:
my asnwer would be yes... as the user above me explaineed it... but if you really want a redundant storage... RAID 1 like the other user wrote.... or RAID 5 , RAID 10 if your willing to spend some extra bucks.... and sorry for my bad english...  :laugh:

kitamesume:
i wonder if its possible to partition a drive to something like 4GB and do software raid-1 with a 4GB ramdisk to make a super fast partition :P

thinking of making fun on it, like moving caches and some games into it =O

SilverDash:
Only specific raid system's and other backup system'ss that actually back the data up to another hardware device are safe. Preferably even to another physical location on the world.

I used to use raid 50 (5+0). That's 6 HD's linked together non-stop backing each other up. Even when 2 disks break you might still have a fully running system. Even your Windows will still run if you pull em out. The problem with raid 5 and raid 50 is that the HD-access is huge. Therefor the HD's will break way sooner than the normally do. HD's weren't designed for that. In my case most HD's didnt even last 1 year... One way to somewhat counter that is by using a 7th HD outside the RAID for the operating system. But I don't recommend RAID 5 or 50 to inexperienced users as there are many things that you might configure wrong.

Another problem with raid systems is that the raid controller itself might break and then you got nothing untill you replace it with the exact same raid controller... And they might be slower in some cases.

But the biggest mother of all problems: If your power supply is bad then every single HD might break. Unless you use an expensive server power supply. Had it only twice in 20 years or so tough. Still sucked. Had to throw away 90% of all hardware including the HD's...


--- Quote ---1)IF A NON-PHYSICAL FAILURE happens, something like that cyclic redundancy or some other "bad data" thing. . . will it fuck up ONLY the partition that its on?
--- End quote ---
Maybe. Depends on the software fail. But almost always the entire drive is gone or random parts all over it.

As for the raid 5 hole.. A few bits on trillions and trillions of bits in total ain't much of a problem I think. Even a single letter uses atleast 8 bits. 1TB = 1000000000000 bytes = 8000000000000 bits. That x 1500 or so and then a few of those might switch. And even when they switch, a bit can only be 0 or 1 so there is still a 50% chance that it is the right one lol.
You can use a UPS system to protect your hardware against power spikes and power outages tough. A bit overkill but it is possible.

Pentium100:

--- Quote from: GoGeTa006 on January 26, 2012, 07:40:23 AM ---1)IF A NON-PHYSICAL FAILURE happens, something like that cyclic redundancy or some other "bad data" thing. . . will it fuck up ONLY the partition that its on?
--- End quote ---

CRC errors are caused by two problems:
1) Those sectors became unreadable because the media or the heads are failing.
2) There was a power problem during writing or a bad data cable (more likely with IDE than SATA or SCSI) resulted in wrong/incomplete data being written in the first place.

In case of #1, the whole drive might fail soon. In case of #2, using more than one partition would only help if you access only one partition at a time.

If a drive has bad blocks, use MHDD and run the "ERASE" command, it will overwrite the entire drive and either fix the bad blocks caused by #2 or cause the drive to remap the bad blocks caused by #1. You will have to back up the drive before doing it.

Bad sectors are the most damaging if they occur in the file system area (FAT file systems have a backup copy of the FAT, but for some weird reason NTFS only has a partial copy of the MFT), otherwise it may be that just a single file is damaged.

Proin Drakenzol:
There is no RAID 0, that's striping.

RAID 1 is eh. RAID 5 is the way to go  8)

Navigation

[0] Message Index

[#] Next page

[*] Previous page