|
|
|
|
| Author |
Message |
Al
Guest
|
 Posted: Tue Jun 05, 2007 6:52 pm Post subject: HD from different manufacturers in a RAID 5 |
 |
|
The latest discussion on this topic I could find in over 4 years old,
so I hope to get some new opinions. I have a RAID 5 array composed of
3 Maxtor Atlas 10K IV 36GB U320 SCA 80-PIN LVD Hard Drives. My RAID
controller is an IBM serveRAID 4Mx. everything is working great.
Normally, we keep spare drives in case one should ever fail, so we can
pull it off the shelf and immediately replace the failed drive.
Unfortunately, Maxtor drives are no longer manufactured, ever since
they were bought out by Seagate. We're running low on spares, and I
want to get some more. I called SeaGate and asked them what drive
matches these specs, and they told me the Chetah 36GB (P/N ST336607LC)
is a perfect match. When I asked if there were any compatibility
issues considering it will go into an existing RAID 5 array made up of
Maxtor drives, I was told "there *shouldn't* be any." Well, I don't
want to take his word for it, he's a salesman - not a RAID expert.
Can someone give me some specific pointers on what to check so I can
be as sure as possible this will work? Or would you recommend I
replace all 3 Maxtors with Cheetah drives? I am quite new to RAID, so
please assume I am an idiot on the subject. Thanks.
Al |
|
| Back to top |
|
 |
Fix your Windows Problems - FAST.
FREE Safe Scan Registry Check. Locate & Fix Errors in Minutes!
|
|
Robert Heller
Guest
|
| Posted: Tue Jun 05, 2007 8:08 pm Post subject: Re: HD from different manufacturers in a RAID 5 |
|
|
At Tue, 05 Jun 2007 13:52:11 -0000 Al <bernala@gmail.com> wrote:
| Quote: |
The latest discussion on this topic I could find in over 4 years old,
so I hope to get some new opinions. I have a RAID 5 array composed of
3 Maxtor Atlas 10K IV 36GB U320 SCA 80-PIN LVD Hard Drives. My RAID
controller is an IBM serveRAID 4Mx. everything is working great.
Normally, we keep spare drives in case one should ever fail, so we can
pull it off the shelf and immediately replace the failed drive.
Unfortunately, Maxtor drives are no longer manufactured, ever since
they were bought out by Seagate. We're running low on spares, and I
want to get some more. I called SeaGate and asked them what drive
matches these specs, and they told me the Chetah 36GB (P/N ST336607LC)
is a perfect match. When I asked if there were any compatibility
issues considering it will go into an existing RAID 5 array made up of
Maxtor drives, I was told "there *shouldn't* be any." Well, I don't
want to take his word for it, he's a salesman - not a RAID expert.
Can someone give me some specific pointers on what to check so I can
be as sure as possible this will work? Or would you recommend I
replace all 3 Maxtors with Cheetah drives? I am quite new to RAID, so
please assume I am an idiot on the subject. Thanks.
Al
|
Below is the status of a perfectly working RAID-5 system, with 6 IBM
148gig and 2 Fujitsu 148gig drives, all but one bundled into a RAID-5
array (one of the Fujitsu is a 'hot spare' (Standby) drive). So yes,
the salesman is correct. The controller should not care about the 'vendor',
so long as the the block count matches. At *worst* if the new disks are
a few blocks larger, it might just not use the extra blocks. I would
just stock up on a few Cheetah drives and replace the Maxtors as they
fail, unless the Maxtors are close to EOL (how old are they?).
***** DAC960 RAID Driver Version 2.4.11 of 11 October 2001 *****
Copyright 1998-2001 by Leonard N. Zubkoff <lnz@dandelion.com>
Configuring Mylex DAC1164P PCI RAID Controller
Firmware Version: 5.08-0-87, Channels: 2, Memory Size: 32MB
PCI Bus: 6, Device: 8, Function: 0, I/O Address: Unassigned
PCI Address: 0xFDBFFC00 mapped at 0xF8844C00, IRQ Channel: 48
Controller Queue Depth: 128, Maximum Blocks per Command: 128
Driver Queue Depth: 127, Scatter/Gather Limit: 33 of 33 Segments
Stripe Size: 64KB, Segment Size: 8KB, BIOS Geometry: 255/63
SAF-TE Enclosure Management Enabled
Physical Devices:
0:0 Vendor: FUJITSU Model: MAP3147NC Revision: 5608
Serial Number: UP36P3900F2V
Disk Status: Online, 286748672 blocks
0:1 Vendor: FUJITSU Model: MAP3147NC Revision: 5608
Serial Number: UP36P3900F85
Disk Status: Standby, 286748672 blocks
0:6 Vendor: PE/PV Model: 1x2 SCSI BP Revision: 1.1
0:7 Vendor: MYLEX Model: DAC1164P Revision: 0508
Serial Number:
1:0 Vendor: IBM Model: IC35L146UCDY10-0 Revision: S27F
Serial Number: ECVDW91C
Disk Status: Online, 286748672 blocks
1:1 Vendor: IBM Model: IC35L146UCDY10-0 Revision: S27F
Serial Number: ECVDAJ6C
Disk Status: Online, 286748672 blocks
1:2 Vendor: IBM Model: IC35L146UCDY10-0 Revision: S27F
Serial Number: ECVD96NC
Disk Status: Online, 286748672 blocks
1:3 Vendor: IBM Model: IC35L146UCDY10-0 Revision: S27F
Serial Number: ECVE7T7C
Disk Status: Online, 286748672 blocks
1:4 Vendor: IBM Model: IC35L146UCDY10-0 Revision: S27F
Serial Number: ECVBZNXC
Disk Status: Online, 286748672 blocks
1:5 Vendor: IBM Model: IC35L146UCDY10-0 Revision: S27F
Serial Number: ECVE7HWC
Disk Status: Online, 286748672 blocks
1:6 Vendor: PE/PV Model: 1x6 SCSI BP Revision: 1.1
1:7 Vendor: MYLEX Model: DAC1164P Revision: 0508
Serial Number:
Logical Drives:
/dev/rd/c0d0: RAID-5, Online, 37748736 blocks, Write Thru
/dev/rd/c0d1: RAID-5, Online, 1682743296 blocks, Write Thru
No Rebuild or Consistency Check in Progress
--
Robert Heller -- 978-544-6933
Deepwoods Software -- Linux Installation and Administration
http://www.deepsoft.com/ -- Web Hosting, with CGI and Database
heller@deepsoft.com -- Contract Programming: C/C++, Tcl/Tk |
|
| Back to top |
|
|
Folkert Rienstra
Guest
|
| Posted: Wed Jun 06, 2007 1:08 am Post subject: Re: HD from different manufacturers in a RAID 5 |
|
|
"Robert Heller" <heller@deepsoft.com> wrote in message news:10d4c$46657c7f$404a99a1$19278@news.news-service.com
| Quote: | At Tue, 05 Jun 2007 13:52:11 -0000 Al <bernala@gmail.com> wrote:
The latest discussion on this topic I could find in over 4 years old,
so I hope to get some new opinions. I have a RAID 5 array composed of
3 Maxtor Atlas 10K IV 36GB U320 SCA 80-PIN LVD Hard Drives. My RAID
controller is an IBM serveRAID 4Mx. everything is working great.
Normally, we keep spare drives in case one should ever fail, so we can
pull it off the shelf and immediately replace the failed drive.
Unfortunately, Maxtor drives are no longer manufactured, ever since
they were bought out by Seagate.
|
Doesn't necessarily mean they aren't available anymore.
| Quote: | We're running low on spares, and I
want to get some more. I called SeaGate and asked them what drive
matches these specs, and they told me the Chetah 36GB (P/N ST336607LC)
is a perfect match. When I asked if there were any compatibility
issues considering it will go into an existing RAID 5 array made up of
Maxtor drives, I was told "there *shouldn't* be any." Well, I don't
want to take his word for it,
he's a salesman - not a RAID expert.
|
Since when can a salesman not be a RAID expert?
| Quote: |
Can someone give me some specific pointers on what to check so I can
be as sure as possible this will work? Or would you recommend I
replace all 3 Maxtors with Cheetah drives? I am quite new to RAID, so
please assume I am an idiot on the subject. Thanks.
Al
Below is the status of a perfectly working RAID-5 system, with 6 IBM
148gig and 2 Fujitsu 148gig drives, all but one bundled into a RAID-5
array (one of the Fujitsu is a 'hot spare' (Standby) drive). So yes,
the salesman is correct. The controller should not care about the 'vendor',
|
So Heller, where exactly does it say serveRAID 4Mx in there, below.
| Quote: | so long as the the block count matches. At *worst* if the new disks are
a few blocks larger, it might just not use the extra blocks.
|
Wrong Heller, at worst is when the new disks are a few blocks smaller.
Atlas typically has 71,833,095 sectors, Cheetah 71,687,372 sectors.
| Quote: | I would just stock up on a few Cheetah drives and replace the Maxtors
as they fail,
|
Yeah that might be a good time to figure out *why* it doesn't work *when*
it doesn't work.
| Quote: | unless the Maxtors are close to EOL (how old are they?).
***** DAC960 RAID Driver Version 2.4.11 of 11 October 2001 *****
Copyright 1998-2001 by Leonard N. Zubkoff <lnz@dandelion.com
Configuring Mylex DAC1164P PCI RAID Controller
Firmware Version: 5.08-0-87, Channels: 2, Memory Size: 32MB
PCI Bus: 6, Device: 8, Function: 0, I/O Address: Unassigned
PCI Address: 0xFDBFFC00 mapped at 0xF8844C00, IRQ Channel: 48
Controller Queue Depth: 128, Maximum Blocks per Command: 128
Driver Queue Depth: 127, Scatter/Gather Limit: 33 of 33 Segments
Stripe Size: 64KB, Segment Size: 8KB, BIOS Geometry: 255/63
SAF-TE Enclosure Management Enabled
Physical Devices:
0:0 Vendor: FUJITSU Model: MAP3147NC Revision: 5608
Serial Number: UP36P3900F2V
Disk Status: Online, 286748672 blocks
0:1 Vendor: FUJITSU Model: MAP3147NC Revision: 5608
Serial Number: UP36P3900F85
Disk Status: Standby, 286748672 blocks
0:6 Vendor: PE/PV Model: 1x2 SCSI BP Revision: 1.1
0:7 Vendor: MYLEX Model: DAC1164P Revision: 0508
Serial Number:
1:0 Vendor: IBM Model: IC35L146UCDY10-0 Revision: S27F
Serial Number: ECVDW91C
Disk Status: Online, 286748672 blocks
1:1 Vendor: IBM Model: IC35L146UCDY10-0 Revision: S27F
Serial Number: ECVDAJ6C
Disk Status: Online, 286748672 blocks
1:2 Vendor: IBM Model: IC35L146UCDY10-0 Revision: S27F
Serial Number: ECVD96NC
Disk Status: Online, 286748672 blocks
1:3 Vendor: IBM Model: IC35L146UCDY10-0 Revision: S27F
Serial Number: ECVE7T7C
Disk Status: Online, 286748672 blocks
1:4 Vendor: IBM Model: IC35L146UCDY10-0 Revision: S27F
Serial Number: ECVBZNXC
Disk Status: Online, 286748672 blocks
1:5 Vendor: IBM Model: IC35L146UCDY10-0 Revision: S27F
Serial Number: ECVE7HWC
Disk Status: Online, 286748672 blocks
1:6 Vendor: PE/PV Model: 1x6 SCSI BP Revision: 1.1
1:7 Vendor: MYLEX Model: DAC1164P Revision: 0508
Serial Number:
Logical Drives:
/dev/rd/c0d0: RAID-5, Online, 37748736 blocks, Write Thru
/dev/rd/c0d1: RAID-5, Online, 1682743296 blocks, Write Thru
No Rebuild or Consistency Check in Progress |
|
|
| Back to top |
|
|
Robert Heller
Guest
|
| Posted: Wed Jun 06, 2007 2:07 pm Post subject: Re: HD from different manufacturers in a RAID 5 |
|
|
At Wed, 06 Jun 2007 12:25:32 -0000 Al <bernala@gmail.com> wrote:
| Quote: |
Thanks for your responses.
I understand the issue about the number of sectors on a drive - that
is something that can be checked easily enough. Are there any other
drive characteristics that you would recommend I compare before
inserting a cheetah HD into the existing RAID 5 array?
For example, does it matter if the cheetah drive has a different
number of platters and heads? Cache on both drive types is 8MB, so
|
The number of platters and heads does not matter -- the firmware on the
disk itself takes care of translating a LBA to physical
platter/head/sector. The RAID controller only ever deals with LBA. So
long as the range of LBAs match (total # of sectors) things should work.
| Quote: | they match there. Throughput is different... do these figures matter?
|
Throughput differences should not matter -- this will be 'smoothed out' by
the RAID controller's own cache. At *worst* overall thoughput will be
whatever the throughput of the slowest disk. I'm presuming that the
cheetahs are faster than the (older) maxtors, in which case as you
migrate from the older disks to the newer ones, the overall thoughput
will get better (faster).
| Quote: |
Thanks for the help
|
--
Robert Heller -- 978-544-6933
Deepwoods Software -- Linux Installation and Administration
http://www.deepsoft.com/ -- Web Hosting, with CGI and Database
heller@deepsoft.com -- Contract Programming: C/C++, Tcl/Tk |
|
| Back to top |
|
|
Al
Guest
|
| Posted: Wed Jun 06, 2007 2:07 pm Post subject: Re: HD from different manufacturers in a RAID 5 |
|
|
Thanks for your responses.
I understand the issue about the number of sectors on a drive - that
is something that can be checked easily enough. Are there any other
drive characteristics that you would recommend I compare before
inserting a cheetah HD into the existing RAID 5 array?
For example, does it matter if the cheetah drive has a different
number of platters and heads? Cache on both drive types is 8MB, so
they match there. Throughput is different... do these figures matter?
Thanks for the help |
|
| Back to top |
|
|
Fix your Windows Problems - FAST.
FREE Safe Scan Registry Check. Locate & Fix Errors in Minutes!
|
|
Folkert Rienstra
Guest
|
| Posted: Thu Jun 07, 2007 2:16 am Post subject: Re: HD from different manufacturers in a RAID 5 |
|
|
"Al" <bernala@gmail.com> wrote in message news:1181132732.438920.72600@h2g2000hsg.googlegroups.com
| Quote: | Thanks for your responses.
I understand the issue about the number of sectors on a drive -
that is something that can be checked easily enough.
|
Oh? I had quite a bit of trouble finding it.
Seagate have been taken over by Maxtor where documentation is concerned. I had to go to the newsgroups and Ebay to get that info.
| Quote: | Are there any other drive characteristics that you would recommend
I compare
|
I have no idea. I have no experience with RAID controllers, let alone the
ServeRAID ones. I have read the users reference docs swiftly and you
don't get much clues there either.
| Quote: | before inserting a cheetah HD into the existing RAID 5 array?
|
You did notice that the cheetah is smaller, did you?
| Quote: |
For example, does it matter if the cheetah drive has a different
number of platters and heads?
|
Besides the difference in capacity, probably not.
| Quote: | Cache on both drive types is 8MB, so they match there.
Throughput is different... do these figures matter?
|
Probably not either. However, should not is not the same as will not.
What you are looking for is someone with lot's of hands-on experience
with serveRAID controllers and I don't think there are any here.
Why not ask IBM themselves?
| Quote: |
Thanks for the help |
|
|
| Back to top |
|
|
Folkert Rienstra
Guest
|
| Posted: Thu Jun 07, 2007 2:18 am Post subject: Re: HD from different manufacturers in a RAID 5 |
|
|
"Robert Heller" <heller@deepsoft.com> wrote in message news:bd0b4$4666b591$404a99a1$14406@news.news-service.com
| Quote: | At Wed, 06 Jun 2007 12:25:32 -0000 Al <bernala@gmail.com> wrote:
Thanks for your responses.
I understand the issue about the number of sectors on a drive - that
is something that can be checked easily enough. Are there any other
drive characteristics that you would recommend I compare before
inserting a cheetah HD into the existing RAID 5 array?
For example, does it matter if the cheetah drive has a different
number of platters and heads? Cache on both drive types is 8MB, so
The number of platters and heads does not matter -- the firmware
on the disk itself takes care of translating a LBA to physical
platter/head/sector. The RAID controller only ever deals with LBA.
So long as the range of LBAs match (total # of sectors) things should work.
|
Pity the cheetah appears to be smaller.
| Quote: |
they match there. Throughput is different... do these figures matter?
Throughput differences should not matter -- this will be 'smoothed out' by
the RAID controller's own cache.
|
As if throughput needs 'smoothing'.
| Quote: | At *worst* overall thoughput will be whatever the throughput of the slowest disk.
|
Nonsense. For that to happen you need to have your stripe size completely
wrong AND have all IO go to the slowest drive, always. That's not possible.
| Quote: | I'm presuming that the cheetahs are faster than the (older) maxtors,
in which case as you migrate from the older disks to the newer ones,
the overall thoughput will get better (faster).
|
Nope. Not with a striped array.
Like you just said above (without realizing it) the slowest drive sets the
pace: n-1 times the slowest drive, max. The faster drives will have to wait
for the slowest one to complete. To get faster, you have to replace all.
| Quote: |
Thanks for the help |
|
|
| Back to top |
|
|
Al
Guest
|
| Posted: Thu Jun 07, 2007 2:12 pm Post subject: Re: HD from different manufacturers in a RAID 5 |
|
|
| Quote: | Why not ask IBM themselves?
|
I did, and I'm just waiting for their response. For the moment, it
seems that number of sectors is the critical factor. I will be using a
74GB cheetah drive (rather than the 36GB) to ensure I have more than
enough sectors. |
|
| Back to top |
|
|
Folkert Rienstra
Guest
|
| Posted: Thu Jun 07, 2007 8:36 pm Post subject: Re: HD from different manufacturers in a RAID 5 |
|
|
"Al" <bernala@gmail.com> wrote in message news:1181220572.949527.22700@h2g2000hsg.googlegroups.com
| Quote: | Why not ask IBM themselves?
I did, and I'm just waiting for their response. For the moment, it
seems that number of sectors is the critical factor.
I will be using a 74GB cheetah drive (rather than the 36GB) to ensure
I have more than enough sectors.
|
That is a bit over the top, don't you think?
You can make a backup, delete the array, put the Cheetah in and
re-setup the array. Restore your backup. All you loose is ~150MB. |
|
| Back to top |
|
|
Robert Heller
Guest
|
| Posted: Thu Jun 07, 2007 8:59 pm Post subject: Re: HD from different manufacturers in a RAID 5 |
|
|
At Thu, 07 Jun 2007 12:49:32 -0000 Al <bernala@gmail.com> wrote:
| Quote: |
Why not ask IBM themselves?
I did, and I'm just waiting for their response. For the moment, it
seems that number of sectors is the critical factor. I will be using a
74GB cheetah drive (rather than the 36GB) to ensure I have more than
enough sectors.
|
If you have any spare (empty) drive bays *and* plan on using signifently
larger drives, it might make sense to re-organize/rebuild the RAID
array.
You'd create a new array and a new logical disk, do a disk-to-disk
backup from the old array to the new array (old logical disk to new
logical disk), then once you are sure the new array is working good,
remove the old array and make the new array the working disk.
--
Robert Heller -- 978-544-6933
Deepwoods Software -- Linux Installation and Administration
http://www.deepsoft.com/ -- Web Hosting, with CGI and Database
heller@deepsoft.com -- Contract Programming: C/C++, Tcl/Tk |
|
| Back to top |
|
|
|
|
Folkert Rienstra
Guest
|
| Posted: Thu Jun 07, 2007 11:15 pm Post subject: Re: HD from different manufacturers in a RAID 5 |
|
|
"Robert Heller" <heller@deepsoft.com> wrote in message news:90215$46682b72$404a99a1$22828@news.news-service.com
| Quote: | At Thu, 07 Jun 2007 12:49:32 -0000 Al <bernala@gmail.com> wrote:
Why not ask IBM themselves?
I did, and I'm just waiting for their response. For the moment, it
seems that number of sectors is the critical factor. I will be using a
74GB cheetah drive (rather than the 36GB) to ensure I have more than
enough sectors.
If you have any
|
How about 3 of 'm, Heller?
| Quote: | spare (empty) drive bays *and* plan on using signifently larger
drives, it might make sense to re-organize/rebuild the RAID array.
|
Heller, what did you not understand in "a 74GB cheetah drive".
| Quote: |
You'd create a new array and a new logical disk, do a disk-to-disk
backup from the old array to the new array (old logical disk to new
logical disk), then once you are sure the new array is working good,
remove the old array and make the new array the working disk. |
|
|
| Back to top |
|
|
|
|
| |
Watched Topics
Search
Register
Log in to check your private messages
Profile
Log in
Memberlist
Usergroups