* Re: [gentoo-user] [OT] Disk recommendations?
2011-04-10 7:50 ` Peter Humphrey
@ 2011-04-10 11:53 ` Stroller
2011-04-10 13:43 ` Peter Humphrey
2011-04-10 13:50 ` Mark Knecht
` (2 subsequent siblings)
3 siblings, 1 reply; 13+ messages in thread
From: Stroller @ 2011-04-10 11:53 UTC (permalink / raw
To: gentoo-user
On 10/4/2011, at 8:50am, Peter Humphrey wrote:
> ...
> I'm just speculating at the moment, from a dabbler's point of view; what benefits
> would accrue from switching from RAID-1 to RAID-5 or above? And, in particular,
> what are the comparative virtues of the Samsung disks?
In your previous message you mention "adding robustness", I don't think you'd change from RAID1 in that case.
RAID5 is less redundant than RAID1, but offers more space per drive.
Either will continue to run if one drive fails, but RAID5 consists of more drives (each of which has the potential for failure).
RAID1 has 2 disks and offers up to 1/2 redundancy. 1/2 your disks can fail without loss of data.
RAID5 has X disks, where X is more than 2, and offers upto 1/X redundancy. If more than 1 drive fails then your data is toast. This inherently allows for data loss if more than only 1/3 or 1/4 (or less - 1/5 or 1/6 if you have enough drives in your system) fail.
RAID6 needs an extra disk over RAID5 (at least 4 total?), and allows 2/X of them to fail whilst still maintaining data integrity.
I guess that theoretically RAID6 might be more robust than RAID1 but realistically one would probably use RAID1 if the volume is intended to be a fixed size (system volume), RAID5 or RAID6 if you want to be able to easily expand the volume (add an extra drive and store more data simply by expanding the filesystem). Other kinds of RAID (10, 50 &c) may be more suitable if read or write speed is also important for specialist applications, but you say you're only interested in home workstation use, not the datacentre.
Note that I only consider hardware RAID - others may be able to give advice more suited to Linux's software RAID.
I use RAID5 for my TV recordings and DVD rips. There's a famous article claiming RAID5 is worthless considering the size of current hard-drives vs uncorrected error rates (which manufacturers express per million or billion bits). I'm sceptical of the article, but nevertheless I guess I'm starting to get paranoid enough I'd prefer RAID6. Unfortunately my hardware RAID controller doesn't support it, so I guess I'm saved the expense. :/
Stroller.
^ permalink raw reply [flat|nested] 13+ messages in thread* Re: [gentoo-user] [OT] Disk recommendations?
2011-04-10 11:53 ` Stroller
@ 2011-04-10 13:43 ` Peter Humphrey
0 siblings, 0 replies; 13+ messages in thread
From: Peter Humphrey @ 2011-04-10 13:43 UTC (permalink / raw
To: gentoo-user
On Sunday 10 April 2011 12:53:39 Stroller wrote:
> On 10/4/2011, at 8:50am, Peter Humphrey wrote:
> > ...
> > I'm just speculating at the moment, from a dabbler's point of view; what
> > benefits would accrue from switching from RAID-1 to RAID-5 or above?
> > And, in particular, what are the comparative virtues of the Samsung
> > disks?
>
> In your previous message you mention "adding robustness", I don't think
> you'd change from RAID1 in that case.
>
> RAID5 is less redundant than RAID1, but offers more space per drive.
>
> Either will continue to run if one drive fails, but RAID5 consists of more
> drives (each of which has the potential for failure).
>
> RAID1 has 2 disks and offers up to 1/2 redundancy. 1/2 your disks can fail
> without loss of data.
>
> RAID5 has X disks, where X is more than 2, and offers upto 1/X redundancy.
> If more than 1 drive fails then your data is toast. This inherently allows
> for data loss if more than only 1/3 or 1/4 (or less - 1/5 or 1/6 if you
> have enough drives in your system) fail.
>
> RAID6 needs an extra disk over RAID5 (at least 4 total?), and allows 2/X of
> them to fail whilst still maintaining data integrity.
>
> I guess that theoretically RAID6 might be more robust than RAID1 but
> realistically one would probably use RAID1 if the volume is intended to be
> a fixed size (system volume), RAID5 or RAID6 if you want to be able to
> easily expand the volume (add an extra drive and store more data simply by
> expanding the filesystem). Other kinds of RAID (10, 50 &c) may be more
> suitable if read or write speed is also important for specialist
> applications, but you say you're only interested in home workstation use,
> not the datacentre.
>
> Note that I only consider hardware RAID - others may be able to give advice
> more suited to Linux's software RAID.
>
> I use RAID5 for my TV recordings and DVD rips. There's a famous article
> claiming RAID5 is worthless considering the size of current hard-drives vs
> uncorrected error rates (which manufacturers express per million or
> billion bits). I'm sceptical of the article, but nevertheless I guess I'm
> starting to get paranoid enough I'd prefer RAID6. Unfortunately my
> hardware RAID controller doesn't support it, so I guess I'm saved the
> expense. :/
Useful info - many thanks!
--
Rgds
Peter
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [gentoo-user] [OT] Disk recommendations?
2011-04-10 7:50 ` Peter Humphrey
2011-04-10 11:53 ` Stroller
@ 2011-04-10 13:50 ` Mark Knecht
2011-04-10 14:23 ` Peter Humphrey
` (2 more replies)
2011-04-10 14:26 ` Dale
2011-04-11 16:51 ` pk
3 siblings, 3 replies; 13+ messages in thread
From: Mark Knecht @ 2011-04-10 13:50 UTC (permalink / raw
To: gentoo-user
On Sun, Apr 10, 2011 at 12:50 AM, Peter Humphrey
<peter@humphrey.ukfsn.org> wrote:
> On Saturday 09 April 2011 22:01:18 Mark Knecht wrote:
>
>> Are you running a RAID?
>
> Yes; mdadm RAID-1, with LVM on top, as in the Gentoo how-to:
> http://www.gentoo.org/doc/en/gentoo-x86+raid+lvm2-quickinstall.xml
>
>> Are you looking for a little redundancy or a lot of redundancy?
>
> I'm just speculating at the moment, from a dabbler's point of view; what benefits
> would accrue from switching from RAID-1 to RAID-5 or above? And, in particular,
> what are the comparative virtues of the Samsung disks?
>
My understanding is there's nothing more reliable than RAID1. mdadm
allows N-wide RAID1. My RAID1's are currently 3-drive.
Typically the higher RAID numbers are for trading off storage space,
redundancy and in some cases throughput. My 5-drive RAID6 gives me
(again, my understanding) equivalent redundancy to a 3-drive RAID1. I
can lose 2 drives in either RAID before I risk losing everything with
a 3rd drive failure, but I only get the storage of 3-drives. A 5-drive
RAID5 would lose everything with 2 drive failures but gets 4 drives
of storage.
As for Samsung drives I have no experience. However one common problem
I read about again and again is a RAID user who loses 1 drive and
then, while in the process of fixing the RAID, loses a second drive.
Most of us (myself included) buy identical drives all at the same time
from the same vendor. This means all the drives were likely from the
same manufacturing batch and, if they are drives that will fail at all
then the group will likely experience multiple drive failures. The
underlying idea of RAID is that the drives are not likely to fail at
the same time giving us time to fix the array. However, if /dev/sda
fails the chances of /dev/sdb failing is higher if they were built at
the same time in the same plant.
Reading the mdadm list for the last couple of years it seems that many
folks running data centers intentionally buy drives from multiple
manufactures, or drives of different sizes from the same manufacturer,
hoping to lower the chances of multiple failures at the same time.
What I did myself was buy 5 drives initially, 3 from Amazon, 2 from
NewEgg. For spares I then waited 2 months, bought one more drive, and
waited another 2 months and got one more. In my case all my drives are
WD RAID Edition drives which have higher reliability specs than the
commercial drives. (And are more expensive and smaller)
As for hardware RAID the risk I hear about there is that if the
controller itself fails then you need an identical backup controller
or you risk the possibility that you won't be able to recover
anything. I don't know how true that is or whether it's just FUD.
Cheers,
Mark
^ permalink raw reply [flat|nested] 13+ messages in thread* Re: [gentoo-user] [OT] Disk recommendations?
2011-04-10 13:50 ` Mark Knecht
@ 2011-04-10 14:23 ` Peter Humphrey
2011-04-11 3:35 ` Stroller
2011-04-11 8:04 ` Helmut Jarausch
2 siblings, 0 replies; 13+ messages in thread
From: Peter Humphrey @ 2011-04-10 14:23 UTC (permalink / raw
To: gentoo-user
On Sunday 10 April 2011 14:50:59 Mark Knecht wrote:
[...]
More useful info - thanks to you too.
> As for hardware RAID the risk I hear about there is that if the
> controller itself fails then you need an identical backup controller
> or you risk the possibility that you won't be able to recover
> anything. I don't know how true that is or whether it's just FUD.
That sounds painful. Makes me glad I'm a skinflint and stuck to software RAID!
--
Rgds
Peter
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [gentoo-user] [OT] Disk recommendations?
2011-04-10 13:50 ` Mark Knecht
2011-04-10 14:23 ` Peter Humphrey
@ 2011-04-11 3:35 ` Stroller
2011-04-11 8:04 ` Helmut Jarausch
2 siblings, 0 replies; 13+ messages in thread
From: Stroller @ 2011-04-11 3:35 UTC (permalink / raw
To: gentoo-user
On 10/4/2011, at 2:50pm, Mark Knecht wrote:
> ... loses 1 drive and
> then, while in the process of fixing the RAID, loses a second drive.
> Most of us (myself included) buy identical drives all at the same time
> from the same vendor. This means all the drives were likely from the
> same manufacturing batch and, if they are drives that will fail at all
> then the group will likely experience multiple drive failures.
It doesn't make it *likely* that they'll fail simultaneously. It makes it less unlikely.
> The
> underlying idea of RAID is that the drives are not likely to fail at
> the same time giving us time to fix the array. However, if /dev/sda
> fails the chances of /dev/sdb failing is higher if they were built at
> the same time in the same plant.
^ This is a more accurate synopsis.
> Reading the mdadm list for the last couple of years it seems that many
> folks running data centers intentionally buy drives from multiple
> manufactures, or drives of different sizes from the same manufacturer,
> hoping to lower the chances of multiple failures at the same time.
I've found it sometimes quite inconvenient to do this, and whilst I consider it good practice I get the impression a lot of people, perhaps the majority, don't bother (or don't even know they should). I kinda think it's a nice thing to do but not essential - I don't know that the risk of simultaneous failure is increased that significantly. Many high-end servers will be sold off-the-shelf by their manufacturers with consecutively-serialed drives in the RAID array - I don't think this is considered risky enough for Dell or IBM to offer non-matching drives as a purchasing option.
One also has to wonder what the performance implications might be of having three drives in an array with slightly different rotational speeds, spin-up and seek times.
Ultimately, we shouldn't be fully dependent upon RAID for the integrity of our data, anyway. "RAID is not a backup" is the famous saying.
> As for hardware RAID the risk I hear about there is that if the
> controller itself fails then you need an identical backup controller
> or you risk the possibility that you won't be able to recover
> anything. I don't know how true that is or whether it's just FUD.
Generally you just need a similar one.
In the case of 3ware you can connect your drives to any other 3ware controller and it will recognise the array descriptors written at the start of the drive.
I haven't swapped drives between the PERCs (rebadged Adaptec, I think) of Dell 2650s & 2850s, but these machines are now so cheap on the secondhand market anyway, you can afford to have a spare identical one.
I think you're over-estimating the *risk* of being unable to find a RAID controller of the same model. But certainly if you buy a good PCIe SATA card on the secondhand market it will not be cheap to replace in the event of failure, and a bargain may not come up on eBay immediately. So I think you'll certainly be able to recover your data, you may just some inconvenience of having to wait to find a cheap enough card or spend a lot of money buying an obsolete card in a hurry. Ideally, you have a spare in advance or buy hardware RAID under a 5-year warranty (in which case it's replaced next-day by the manufacturer).
This is really a matter of horses-for-courses. Most people (including myself) don't really "need" hardware RAID. Hardware RAID is much more expensive, but I do consider it "better", if only because you can hot-swap. That is not assured with cheap SATA controllers.
OTOH Linux's software RAID does seem to be just as fast (??) as hardware RAID, and has some cool features.
Stroller.
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [gentoo-user] [OT] Disk recommendations?
2011-04-10 13:50 ` Mark Knecht
2011-04-10 14:23 ` Peter Humphrey
2011-04-11 3:35 ` Stroller
@ 2011-04-11 8:04 ` Helmut Jarausch
2 siblings, 0 replies; 13+ messages in thread
From: Helmut Jarausch @ 2011-04-11 8:04 UTC (permalink / raw
To: gentoo-user
On 04/10/2011 03:50:59 PM, Mark Knecht wrote:
> On Sun, Apr 10, 2011 at 12:50 AM, Peter Humphrey
> <peter@humphrey.ukfsn.org> wrote:
> ....
> I read about again and again is a RAID user who loses 1 drive and
> then, while in the process of fixing the RAID, loses a second drive.
> Most of us (myself included) buy identical drives all at the same
> time
> from the same vendor. This means all the drives were likely from the
> same manufacturing batch and, if they are drives that will fail at
> all
> then the group will likely experience multiple drive failures. The
Life time is always a statistical statement and therefore I consider it
very unlikely that two or more of them fail at the same time - perhaps
with one (big) exception. A given brand/model might be very sensitive
to power failures or excess voltages, and in that case many of them
could fail after a power failure - or excess voltage.
Since my data is not rapidly changing, I backup my data several times a
day onto a spare disk which I spin down in between (using hdparm).
At least I hope to increase the life time of that backup disk this way.
Helmut.
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [gentoo-user] [OT] Disk recommendations?
2011-04-10 7:50 ` Peter Humphrey
2011-04-10 11:53 ` Stroller
2011-04-10 13:50 ` Mark Knecht
@ 2011-04-10 14:26 ` Dale
2011-04-11 16:51 ` pk
3 siblings, 0 replies; 13+ messages in thread
From: Dale @ 2011-04-10 14:26 UTC (permalink / raw
To: gentoo-user
Peter Humphrey wrote:
>
> I'm just speculating at the moment, from a dabbler's point of view; what benefits
> would accrue from switching from RAID-1 to RAID-5 or above? And, in particular,
> what are the comparative virtues of the Samsung disks?
>
>
I have one 750Gb Samsung drive that I have had at least a year. So far,
no problems and it is pretty fast. It does what it is supposed to for a
3Gbs/sec drive. That said, it may die next week. We all know how this
works. They work just fine until they die, usually with no warning or
very little warning.
I might also add, I have a UPS and surge protection coming from the
wall, plus surge and voltage protection built into the UPS. I have yet
to have a drive die so maybe all that does help. I dunno.
I have several Western Digital drives, no problems so far although I
don't like the green ones to much.
Just my $0.02 worth.
Dale
:-) :-)
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [gentoo-user] [OT] Disk recommendations?
2011-04-10 7:50 ` Peter Humphrey
` (2 preceding siblings ...)
2011-04-10 14:26 ` Dale
@ 2011-04-11 16:51 ` pk
2011-04-11 17:41 ` Peter Humphrey
3 siblings, 1 reply; 13+ messages in thread
From: pk @ 2011-04-11 16:51 UTC (permalink / raw
To: gentoo-user
On 2011-04-10 09:50, Peter Humphrey wrote:
> what are the comparative virtues of the Samsung disks?
Not sure if you know about this (only affects F4 Ecogreen 2TB):
http://sourceforge.net/apps/trac/smartmontools/wiki/SamsungF4EGBadBlocks
Best regards
Peter K
^ permalink raw reply [flat|nested] 13+ messages in thread