What happens to discs in a JBOD 'raid' array when one crashes.
JBOD isn't RAID but it is mentioned a lot with RAID controllers.
I found the following text on the net:
"A lot of people talk about JBOD and disk spanning as being
considered forms of RAID. This is not really true; JBOD stands
for Just a Bunch Of Disks and its purpose is to combine
multiple hard disks into one. It does not provide redundancy,
and therefore is not technically a level of RAID. However, JBOD
is useful if you have an application that requires mass amounts
of contiguous hard disk space and you don't have a large drive
to use. This should only be used as a band-aid fix though, as
several drives combined leads to increased chances of failure."
So what is the risk if one drive failes?
Do I lose the whole array?
The reason I ask: I am not concerned about data safety.
I have a server just as a data store for a media streamer.
Everything stored on that server is also backed-up on cd's
and dvd's.
But if it means that, if one drive fails it ruins the whole array,
I'd rather not use jbod at all.
(I'm talking about 6TB of data and I don't want to copy
*everything* if 1TB drive failes.)
Re: What happens to discs in a JBOD 'raid' array when one crashes.
JBOD is used where each disk is used for a different purpose e.g. one for each OS, one for data, one for the page file, etc.. They can do double duty e.g. the drive with the OS can be used to backup data files from the disk with the working data. JBOD does not usually involve an array, but you could use software RAID to span disks.
Marcel Overweel wrote:
>
> JBOD isn't RAID but it is mentioned a lot with RAID controllers.
> I found the following text on the net:
>
> "A lot of people talk about JBOD and disk spanning as being
> considered forms of RAID. This is not really true; JBOD stands
> for Just a Bunch Of Disks and its purpose is to combine
> multiple hard disks into one. It does not provide redundancy,
> and therefore is not technically a level of RAID. However, JBOD
> is useful if you have an application that requires mass amounts
> of contiguous hard disk space and you don't have a large drive
> to use. This should only be used as a band-aid fix though, as
> several drives combined leads to increased chances of failure."
>
> So what is the risk if one drive failes?
> Do I lose the whole array?
>
> The reason I ask: I am not concerned about data safety.
> I have a server just as a data store for a media streamer.
> Everything stored on that server is also backed-up on cd's
> and dvd's.
>
> But if it means that, if one drive fails it ruins the whole array,
> I'd rather not use jbod at all.
> (I'm talking about 6TB of data and I don't want to copy
> *everything* if 1TB drive failes.)
>
> regards,
> Marcel
>JBOD isn't RAID but it is mentioned a lot with RAID controllers.
>I found the following text on the net:
>
>"A lot of people talk about JBOD and disk spanning as being
>considered forms of RAID. This is not really true; JBOD stands
>for Just a Bunch Of Disks and its purpose is to combine
>multiple hard disks into one.
That would be a multi-drive span. You can do that with
typical raid controllers, OR instead you could do single
drive spans which is the same as what a non-raid controller
on any motherboard does, sometimes with the exception of
writing metadata which is only used for identification
purposes, typically with single drive spans it just
identifies it is an existing volume, not a newly added drive
that should be flagged for input from the system
administrator for some designation of purpose for the
drive... and yet typically a newly added drive that is not
designated at all, defaults to single drive span. Just
don't manually set it to single drive stripe as this makes
it's access proprietary to the raid controller used to set
it up (or one intercompatible with that controller which is
still fairly limiting).
>It does not provide redundancy,
>and therefore is not technically a level of RAID. However, JBOD
>is useful if you have an application that requires mass amounts
>of contiguous hard disk space and you don't have a large drive
>to use. This should only be used as a band-aid fix though, as
>several drives combined leads to increased chances of failure."
>
>So what is the risk if one drive failes?
>Do I lose the whole array?
>
>The reason I ask: I am not concerned about data safety.
>I have a server just as a data store for a media streamer.
>Everything stored on that server is also backed-up on cd's
>and dvd's.
>
>But if it means that, if one drive fails it ruins the whole array,
>I'd rather not use jbod at all.
>(I'm talking about 6TB of data and I don't want to copy
>*everything* if 1TB drive failes.)
6TB is way too much to be backing up with DVDs, IMO, let
alone CDs. Consider either tape or hard drives for backup.
If one drive fails I do think you lose the whole array in
practical terms of it being readily accessible for use, but
at least unlike RAID0, there is in theory a good chance a
disk scanning type of recovery software could salvage some
if not all of the data remaining on the viable drives. I
wrote "in theory" because I have never recovered data from a
JBOD, would never use one. You should do a test by copying
test data, actually unplugging a member then seeing if you
can recover from it in a timely fashion.
In the case of wanting a lot of hard to manage (for recovery
due to size) data intact I would use RAID5 instead. That
incurrs the cost of a more elaborate RAID controller if
yours doesn't support it, and needing one addt'l drive
compared to what you would've needed for JBOD, but having
the ability to recover gracefully from a single drive
failure makes it seem the best option in your situation.
Re: What happens to discs in a JBOD 'raid' array when one crashes.
"kony" <spam@spam.com> schreef in bericht
newsjr4741p946mtt661rgid18v0krjr03oq5@4ax.com...
> On Mon, 7 Jul 2008 12:55:36 +0200, "Marcel Overweel"
> <moverweel@hotmail.com> wrote:
>
>>JBOD isn't RAID but it is mentioned a lot with RAID controllers.
>>I found the following text on the net:
>>
>>"A lot of people talk about JBOD and disk spanning as being
>>considered forms of RAID. This is not really true; JBOD stands
>>for Just a Bunch Of Disks and its purpose is to combine
>>multiple hard disks into one.
>
> That would be a multi-drive span. You can do that with
> typical raid controllers, OR instead you could do single
> drive spans which is the same as what a non-raid controller
> on any motherboard does, sometimes with the exception of
> writing metadata which is only used for identification
> purposes, typically with single drive spans it just
> identifies it is an existing volume, not a newly added drive
> that should be flagged for input from the system
> administrator for some designation of purpose for the
> drive... and yet typically a newly added drive that is not
> designated at all, defaults to single drive span. Just
> don't manually set it to single drive stripe as this makes
> it's access proprietary to the raid controller used to set
> it up (or one intercompatible with that controller which is
> still fairly limiting).
>
>
>
>>It does not provide redundancy,
>>and therefore is not technically a level of RAID. However, JBOD
>>is useful if you have an application that requires mass amounts
>>of contiguous hard disk space and you don't have a large drive
>>to use. This should only be used as a band-aid fix though, as
>>several drives combined leads to increased chances of failure."
>>
>>So what is the risk if one drive failes?
>>Do I lose the whole array?
>>
>>The reason I ask: I am not concerned about data safety.
>>I have a server just as a data store for a media streamer.
>>Everything stored on that server is also backed-up on cd's
>>and dvd's.
>>
>>But if it means that, if one drive fails it ruins the whole array,
>>I'd rather not use jbod at all.
>>(I'm talking about 6TB of data and I don't want to copy
>>*everything* if 1TB drive failes.)
>
>
> 6TB is way too much to be backing up with DVDs, IMO, let
> alone CDs. Consider either tape or hard drives for backup.
>
I think I used the wrong words here, this server is for
mediastreaming only, so all that's going to be put on the
drives is already stored somewhere else.
Mostly on DVD's.
> If one drive fails I do think you lose the whole array in
> practical terms of it being readily accessible for use, but
> at least unlike RAID0, there is in theory a good chance a
> disk scanning type of recovery software could salvage some
> if not all of the data remaining on the viable drives. I
> wrote "in theory" because I have never recovered data from a
> JBOD, would never use one. You should do a test by copying
> test data, actually unplugging a member then seeing if you
> can recover from it in a timely fashion.
Yeah, that's a good idea. I can copy a huge zip or rar file over
and over again each with a different name and see what
happens if one drive gets unplugged.
> In the case of wanting a lot of hard to manage (for recovery
> due to size) data intact I would use RAID5 instead. That
> incurrs the cost of a more elaborate RAID controller if
> yours doesn't support it, and needing one addt'l drive
> compared to what you would've needed for JBOD, but having
> the ability to recover gracefully from a single drive
> failure makes it seem the best option in your situation.
I will look into that as well. I'm fairly new with raid and jbod
did sound very interesting because for the mediastream
it is impossible to point to more than four sources.
Re: What happens to discs in a JBOD 'raid' array when one crashes.
ok, thanks!
regards,
Marcel
"Mike Walsh" <spam_sucks@bellsouth.net> schreef in bericht
news:487236FD.3DB71E66@bellsouth.net...
>
> JBOD is used where each disk is used for a different purpose e.g. one for
> each OS, one for data, one for the page file, etc.. They can do double
> duty e.g. the drive with the OS can be used to backup data files from the
> disk with the working data. JBOD does not usually involve an array, but
> you could use software RAID to span disks.
>
> Marcel Overweel wrote:
>>
>> JBOD isn't RAID but it is mentioned a lot with RAID controllers.
>> I found the following text on the net:
>>
>> "A lot of people talk about JBOD and disk spanning as being
>> considered forms of RAID. This is not really true; JBOD stands
>> for Just a Bunch Of Disks and its purpose is to combine
>> multiple hard disks into one. It does not provide redundancy,
>> and therefore is not technically a level of RAID. However, JBOD
>> is useful if you have an application that requires mass amounts
>> of contiguous hard disk space and you don't have a large drive
>> to use. This should only be used as a band-aid fix though, as
>> several drives combined leads to increased chances of failure."
>>
>> So what is the risk if one drive failes?
>> Do I lose the whole array?
>>
>> The reason I ask: I am not concerned about data safety.
>> I have a server just as a data store for a media streamer.
>> Everything stored on that server is also backed-up on cd's
>> and dvd's.
>>
>> But if it means that, if one drive fails it ruins the whole array,
>> I'd rather not use jbod at all.
>> (I'm talking about 6TB of data and I don't want to copy
>> *everything* if 1TB drive failes.)
>>
>> regards,
>> Marcel
>
> --
> Mike Walsh
>> That would be a multi-drive span. You can do that with
>> typical raid controllers, OR instead you could do single
>> drive spans which is the same as what a non-raid controller
>> on any motherboard does, sometimes with the exception of
>> writing metadata which is only used for identification
>> purposes, typically with single drive spans it just
>> identifies it is an existing volume, not a newly added drive
>> that should be flagged for input from the system
>> administrator for some designation of purpose for the
>> drive... and yet typically a newly added drive that is not
>> designated at all, defaults to single drive span. Just
>> don't manually set it to single drive stripe as this makes
>> it's access proprietary to the raid controller used to set
>> it up (or one intercompatible with that controller which is
>> still fairly limiting).
>
>I think I used the wrong words here, this server is for
>mediastreaming only, so all that's going to be put on the
>drives is already stored somewhere else.
>Mostly on DVD's.
Understood, but I failed to continue a thought I had started
above. You could do a multi-drive span to have one logical
volume at the raid controller level, but instead you can
also do several independant single drive spans then map them
to one at the OS level, thereby doing away with the issue
you were concerned about, only needing to restore the
contents of one drive/volume in the event of one failing.
We don't know the importance of uptime though, if this is
business related and ALL content needs to have continual
availability, it sways towards at least a bit of RAID
redundancy being a good idea, like the RAID5 I mentioned
previously, possibly even a hot spare already installed and
ready for a rebuild.
>
>> If one drive fails I do think you lose the whole array in
>> practical terms of it being readily accessible for use, but
>> at least unlike RAID0, there is in theory a good chance a
>> disk scanning type of recovery software could salvage some
>> if not all of the data remaining on the viable drives. I
>> wrote "in theory" because I have never recovered data from a
>> JBOD, would never use one. You should do a test by copying
>> test data, actually unplugging a member then seeing if you
>> can recover from it in a timely fashion.
>
>Yeah, that's a good idea. I can copy a huge zip or rar file over
>and over again each with a different name and see what
>happens if one drive gets unplugged.
>
>> In the case of wanting a lot of hard to manage (for recovery
>> due to size) data intact I would use RAID5 instead. That
>> incurrs the cost of a more elaborate RAID controller if
>> yours doesn't support it, and needing one addt'l drive
>> compared to what you would've needed for JBOD, but having
>> the ability to recover gracefully from a single drive
>> failure makes it seem the best option in your situation.
>
>I will look into that as well. I'm fairly new with raid and jbod
>did sound very interesting because for the mediastream
>it is impossible to point to more than four sources.
I'd do the RAID5 or map the single drive spans to one volume
via OS, before a multi-drive span.