[MLUG] JBOD
Jeremy Chapman
me at jeremychapman.info
Fri Mar 14 15:42:44 EDT 2008
Patrick McLean wrote:
> Jeremy Chapman wrote:
>
>> If a drive fails in a JBOD array, can data still be read? Do files get
>> sent anywhere to either disk, or is there an attempt made to keep data
>> together on one disk?
>>
>>
>
> JBOD stands for "Just a Bunch Of Disks" it not RAID, it just means all
> your disks are treated as individual disks. The name is there as a way
> to specify that you don't want to use the RAID features of your disk
> controller. As such there is no redundancy, all disks have individual,
> separate OS-visible devices, and all data stored on the filesystem on a
> disk is only stored on that disk.
I am talking about old school JBOD, concatenation of disks, not what
some RAID cards call non-raided disks. This is confusing hehe :) What I
mean is like LVM in that different disks can be put together to appear
as one continuous disk.
The answer to my original question seems to be that data is written
sequentially to first disk until full, then spans over to next disk. It
appears then that only a spanned file would be lost if a disk fails (of
course with everything on failing drive as well). It does not appear to
corrupt entire array. ie http://en.wikipedia.org/wiki/Image:JBOD.svg
While not really a RAID level, it is usually implemented by raid
controllers afaik - check here
http://en.wikipedia.org/wiki/Standard_RAID_levels
This is basically what I was talking about "Many Linux
<http://en.wikipedia.org/wiki/Linux> distributions use the terms "linear
mode" or "append mode". The Mac OS X
<http://en.wikipedia.org/wiki/Mac_OS_X> 10.4 implementation — called a
"Concatenated Disk Set" — does not leave the user with any usable data
on the remaining drives if one drive fails in a concatenated disk set,
although the disks otherwise operate as described above." - Mac file
system would fail in the way I was afraid of.
Jeremy
More information about the mlug
mailing list