|
Home > Archive > Data Storage > January 2006 > What is an individual disk drive failure rate of HDS storage systems?
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
| Author |
What is an individual disk drive failure rate of HDS storage systems?
|
|
| hellur 2006-01-14, 2:46 am |
| Hello,
I'm considering buying a new enterprise storage system.
I know there are a lot of factors to be considered, and most of them
are reviewed.
But one of the factors which is not clear yet is the individual disk
drive failure rate.
It may not be important as long as the system is protected by proper
raid configurations, but I've still got this request by the company to
adapt storage systems, and at least I need to explain the company with
proper reason.
Is the individual disk drive failure rate of HDS storage systems lower
than that of EMC?
I heard the disk drive failure rate of EMC storages is a lot higher
than that of HDS.
But others provide some reasons, which may make sense in some way, for
example, HDS leaves disk drives until they are really "destroyed", but
EMC replaces them proactively before they encounter serious problems.
I'm not quite sure whether above statements are true or not.
Can anyone provide me any idea or data if HDS systems' drive failure
rate is really higher, and why?
| |
| John S. 2006-01-14, 2:46 am |
| We have two HDS 9980Vs.
The first one has 136 disks. Last year it looks like this array had 6
disk replaced.
The second array has 160 disks and it had 5 disks replaced.
I know that HDS proactivly replaces disks... I don't know how many of
the 11 disks that were replaced were "dead" and how many were simply
showing signs of a pending failure...
These 2 arrays are also 2.5 years old...
| |
|
| I'm currently using a HDS Thunder 9570. The storage subsystem does a
pre-emptive call back when it detects hard disk abnormality.
My array is still in working condition(all disks showing activity).
When the HDS guy came to swap the disk. Only the spare, where the data
had been copied to and the replaced disk are busy(having lots of
activity). The whole disk group was not rebuilt.
Well, If you're considering HDS, i would encourage you to have a look
at their NSC55. Read the ESG's report. It's truly remarkable. Hot
online migrating of Database to another storage subsystem with no
downtime. In one of the test, they even show that a client can
streaming video from the storage without interruption while migrating
to another subsystem. This is what i call true virtualization. I would
love tiered storage for myself. As not all data is made equal, I would
love to transfer them to a lower end storage array without getting the
data offline. You could use any lower end storage and virtualize them
with the NSC55. Including EMCs.... HP's MSA
| |
|
| To add on, If you are so afraid of having a raid group failure, choose
a storage subsystem with RAID6 capability. Especially important if you
want to use SATA disks. Imagine the prolong disk activity when building
8X400GB of cheap unreliable disks. It may fail anytime when u're
rebuilding it.
I've had SATA array rebuild failure(Especially dangerous on Huge array
of cheap unreliable disks) on my direct attach storage(Dell servers)
before. Trusts me... You don't wanna go there.
| |
| Faeandar 2006-01-20, 5:48 pm |
| On 12 Jan 2006 00:52:26 -0800, "hellur" <qkaths@hanmail.net> wrote:
>Hello,
>I'm considering buying a new enterprise storage system.
>I know there are a lot of factors to be considered, and most of them
>are reviewed.
>But one of the factors which is not clear yet is the individual disk
>drive failure rate.
>It may not be important as long as the system is protected by proper
>raid configurations, but I've still got this request by the company to
>adapt storage systems, and at least I need to explain the company with
>proper reason.
>
>Is the individual disk drive failure rate of HDS storage systems lower
>than that of EMC?
>I heard the disk drive failure rate of EMC storages is a lot higher
>than that of HDS.
>But others provide some reasons, which may make sense in some way, for
>example, HDS leaves disk drives until they are really "destroyed", but
>EMC replaces them proactively before they encounter serious problems.
>
>I'm not quite sure whether above statements are true or not.
>Can anyone provide me any idea or data if HDS systems' drive failure
>rate is really higher, and why?
They use the same drives from the same manufacturers. Drive failure
rate should hardly be a factor in your decision.
~F
| |
| Stuart 2006-01-21, 2:46 am |
| hi,
all arrays suffer from drive failures and most use the same drive
vendor to purchase their disks.... failures prob average out between
say emc / hds / netapp / 3part / eva etc... the actual question of
drive failure becomes irrelevent.
What you should really be asking is how the pre-emtive diags work out
when a drive fails, and more importantly what drive rebuild times are.
As drives become bigger in density, drive rebuild could and will start
stretching out - this becomes the biggest question. focus your
questions to the vendor on how RAID is employed, Understanding how the
array diags behave, what cpu hit you may get within the array if you
are calculating data from checksum information, and also - not just
how raid is employed from a complete disk level but also you may use
drive splits so that data from one individual disk (and each of its
splits) may have a raid-1 mirror to a whole bunch of other disks.
anyway - that is a few of my ideas
all of these vendors are offering decent electronics and "brown
spinney stuff" it is how they differentiate themselves.
Also - dont forget the most important thing - how good their support
model is... drives fail and get replaced - but if you dont get the
engineer to site to fix this stuff, and is good at his / her "stuff"
you will really struggle..
hope this helps,
S.
On 12 Jan 2006 00:52:26 -0800, "hellur" <qkaths@hanmail.net> wrote:
>Hello,
>I'm considering buying a new enterprise storage system.
>I know there are a lot of factors to be considered, and most of them
>are reviewed.
>But one of the factors which is not clear yet is the individual disk
>drive failure rate.
>It may not be important as long as the system is protected by proper
>raid configurations, but I've still got this request by the company to
>adapt storage systems, and at least I need to explain the company with
>proper reason.
>
>Is the individual disk drive failure rate of HDS storage systems lower
>than that of EMC?
>I heard the disk drive failure rate of EMC storages is a lot higher
>than that of HDS.
>But others provide some reasons, which may make sense in some way, for
>example, HDS leaves disk drives until they are really "destroyed", but
>EMC replaces them proactively before they encounter serious problems.
>
>I'm not quite sure whether above statements are true or not.
>Can anyone provide me any idea or data if HDS systems' drive failure
>rate is really higher, and why?
| |
| Stuart 2006-01-21, 2:46 am |
| PS - dont believe all of the gossip you hear of drive failures from
EMC - track record we have had with them has been excellent... we have
literally hundereds of symms within our org - and they are very well
behaved.
On 12 Jan 2006 00:52:26 -0800, "hellur" <qkaths@hanmail.net> wrote:
>Hello,
>I'm considering buying a new enterprise storage system.
>I know there are a lot of factors to be considered, and most of them
>are reviewed.
>But one of the factors which is not clear yet is the individual disk
>drive failure rate.
>It may not be important as long as the system is protected by proper
>raid configurations, but I've still got this request by the company to
>adapt storage systems, and at least I need to explain the company with
>proper reason.
>
>Is the individual disk drive failure rate of HDS storage systems lower
>than that of EMC?
>I heard the disk drive failure rate of EMC storages is a lot higher
>than that of HDS.
>But others provide some reasons, which may make sense in some way, for
>example, HDS leaves disk drives until they are really "destroyed", but
>EMC replaces them proactively before they encounter serious problems.
>
>I'm not quite sure whether above statements are true or not.
>Can anyone provide me any idea or data if HDS systems' drive failure
>rate is really higher, and why?
| |
| John S. 2006-01-23, 7:48 am |
| Actually HDS makes thier own drives... and claim a higher MTBF...
FWIW...
| |
|
| On 23 Jan 2006 03:33:32 -0800, "John S." wrote:
>Actually HDS makes thier own drives... and claim a higher MTBF...
>FWIW...
HDS don't always use their own drives... ;)
HVB.
| |
| Scott Howard 2006-01-23, 7:48 am |
| John S. <fishinjts@netscape.net> wrote:
> Actually HDS makes thier own drives... and claim a higher MTBF...
Whilst Hitachi do make their own drives, the MTBF's are basically the
same as the other equivalent drives from other manufacturers. ie, within
the 1.0 to 1.5 million hours range.
Scott.
| |
| Scott Howard 2006-01-23, 7:48 am |
| hellur <qkaths@hanmail.net> wrote:
> Is the individual disk drive failure rate of HDS storage systems lower
> than that of EMC?
It depends how you measure "failure". Both EMC and HDS arrays will have
disks "replaced" at a higher rate than most other arrays, and in fact in
many cases at higher than MTBF rates.
But the majority of the disks will not have "failed" as such - they will
actually be proactive replacements, where the array has detected one of a
number of indicators that the disk _may_ be on it's way to failure, and
will hot-spare it out before it actually fails.
The nett result is an apparent higher rate of failures, but also a lower
chance of actual data loss as the replacements are proactive (pre-failure)
rather than reactive (post-failure).
Scott
|
|
|
|
|