Overclock.net › Forums › Components › Hard Drives & Storage › RAID Controllers and Software › PERC 5/i RAID Card: Tips and Benchmarks
New Posts  All Forums:Forum Nav:

PERC 5/i RAID Card: Tips and Benchmarks - Page 457

post #4561 of 7150
Quote:
Originally Posted by diehardfan View Post
Here are some of my questions/concern:

-I understand most of you are using RAID 5, which I intend to use too but BER (Bit-Error-Rate) was bought to my attention (More info on BER here 1, 2, 3). Since Percs has BBU does this mitigate the problem and gives you guys the confidence in RAID 5 using Percs?
No, the BBU does nothing for BER issues, also known as UREs (Unrecoverable Read Error). There is "some" debate among storage professionals in the IT world about the reality of this problem; or whether one should really be concerned about it or of it is just a theoretical problem rarely encountered. I will tell you from my experience setting up storage systems for large businesses, that UREs are _real_ problems. I run into them at least a handful of times every year; meaning, I'm dealing with a URE at least once every few months. At this point, I caution all my clients about this issue if they are using 500GB or larger drives in arrays that include more than 7 disks. This threshold is empirical, that is, that's about the smallest array that I've had encounters with UREs.

So, then the question is, what can you do about it? A couple of things:

1) if your data is important to you, make sure you have an adequate backup routine.
2) use RAID-6 instead of RAID-5.
3) make sure you run a consistency check at least once a month. if the URE is found during the consistency check, it can rebuild the damaged stripe using parity and upon re-write, the bad block will get reallocated.
4) know how to recover from a URE problem. in most cases, i run into UREs during a degraded rebuild of RAID-5... (yes, literally the worst times to encounter them) the blocks that cannot be read are lost, but you can recovery everything else and hopefully recover the lost blocks from backups or some other means.
5) if you have an operating system that supports block level error checking (checksums), use it to detect bit rot. something like ZFS in Solaris.

Quote:
Originally Posted by diehardfan View Post
-I am sure most of you have heard about TLER/CCTL/ERC and the importance of it when using RAID 5 (More info here, TLER). It seems my 1TB WD Black cannot have their TLER enabled. Should I be concerned? Should I get drives with TLER/CCTL/ERC enabled?
If you want a stable RAID-5, then use drives with TLER or similar feature. Or, just use really old drives where error recovery wasn't suppose to take so long. ;-P I personally use WD RE3/ RE4 drives... If you don't, every so often, you might notice a drive drop out of the array. That could be rather inconvenient.. but if you're one who likes to drink from a fire hose, it might make you learn really quick how to deal with degraded or offline RAID arrays. ;-P

Quote:
Originally Posted by diehardfan View Post
-Since Perc can handle 8 SATA drives in total. I want to build one RAID drive using 4 X 1TB and 4 X 250GB for two seperate RAID drives. Can either Perc 5 or 6 do it? Or would you recommend combining all 8 drives into on RAID drive?

Thanks
your proposed scenario is possible. totally normal and somewhat typical. but, if you want maximum performance, then you should allocated all drives to a single RAID array. the #1 performance factor in RAID is always spindle count...
Edited by BLinux - 9/15/10 at 8:24pm
TAIPEI
(10 items)
 
AURORA
(13 items)
 
 
MotherboardGraphicsRAMHard Drive
ASRock X99 Extreme11 EVGA GTX 980 Superclocked 32GB 8x4GB Corsair LPX Samsung XP941  
Hard DriveCoolingOSMonitor
Western Digital 3TB RE Noctua NH-D15 Fedora 21 Linux Samsung S27D590C 
PowerCase
Seasonic SS-1200XP Cooler Master Cosmos II 
CPUMotherboardGraphicsRAM
Dual Quad-core L5430 2.66Ghz 12mb cache Intel 5000 chipset ATI ES1000 64GB FBDIMM DDR2 PC2-5300 667Mhz 
Hard DriveOSPower
WD3000FYYZ PERC H700 w/ 512MB cache CentOS 7.2.1511 950W x2 
  hide details  
Reply
TAIPEI
(10 items)
 
AURORA
(13 items)
 
 
MotherboardGraphicsRAMHard Drive
ASRock X99 Extreme11 EVGA GTX 980 Superclocked 32GB 8x4GB Corsair LPX Samsung XP941  
Hard DriveCoolingOSMonitor
Western Digital 3TB RE Noctua NH-D15 Fedora 21 Linux Samsung S27D590C 
PowerCase
Seasonic SS-1200XP Cooler Master Cosmos II 
CPUMotherboardGraphicsRAM
Dual Quad-core L5430 2.66Ghz 12mb cache Intel 5000 chipset ATI ES1000 64GB FBDIMM DDR2 PC2-5300 667Mhz 
Hard DriveOSPower
WD3000FYYZ PERC H700 w/ 512MB cache CentOS 7.2.1511 950W x2 
  hide details  
Reply
post #4562 of 7150
While I agree with the statements above about WDTLER and UREs, I'll just give my own firsthand account:

I'm using a Dell PERC 5/i running the final Dell firmware, with 7x WD1001FALS 1TB Black drives. Two of them were old enough to have WDTLER work, the other 5 are too new and the firmware on them blocks it. I'm using a RAID5 array, and I have 5.45 TB usable space accordingly. I have consumed 60% of that storage thusfar, mostly with backups from the systems in my house, and a good amount of shared media. I have made a point to run an archive of steam games for rapid reinstallation

I have been running my array for about 9 months without a single problem. I have the BBU module, and have both read and write caching enabled. I also have the entire system on a UPS, in a rackmount chassis, mounted in a rack in the basement, which I have leveled to the concrete floor. The whole setup weighs about 300 Lbs. The server is running Windows Server 2008 R2. I'm pleased as punch.

I acknowledge that if I had a failure, I could lose all my data, and that I would be really sad, possibly miserable. Thing is, it's just data, and life goes on. More importantly, the array is stable, and I have no fear of losing data. The chassis has plenty of intake, and even in the hot summer, the basement stays cool. The system has never once overheated, the PERC 5/i has a nice cooling solution I concocted, and I submit that the likelihood of data loss to be extremely small.

Having said all of that, when I replace my array, I'll be going with a PERC 6/i, RAID6, and RE4 drives (or better, whatever is available). I'm going to use 2TB next time. Just waiting for the price to be right.

TL;DR: I'm running drives that *could* suffer from URE, but I haven't had a problem yet. However, I'm aware of it. The point is to be aware; to know the risks.
ROSTZWEI
(17 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i7-980X ASUS RAMPAGE III FORMULA XFX HD6970 2GB 24GB DDR3-1600 (G.Skill RipJaws) 
Hard DriveCoolingOSMonitor
WD6000HLHX (Western Digital Velociraptor 600GB) Noctua NH-D14 Windows 7 Ultimate x64 HP LP2475w 
MonitorMonitorKeyboardPower
HP ZR24w HP ZR24w Rosewill RK-9000BR Antec CP-850 
CaseMouseMouse PadAudio
Antec P183 Logitech G700 Razer Goliathus Extended Speed Edition Auzentech HomeTheater HD 
  hide details  
Reply
ROSTZWEI
(17 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i7-980X ASUS RAMPAGE III FORMULA XFX HD6970 2GB 24GB DDR3-1600 (G.Skill RipJaws) 
Hard DriveCoolingOSMonitor
WD6000HLHX (Western Digital Velociraptor 600GB) Noctua NH-D14 Windows 7 Ultimate x64 HP LP2475w 
MonitorMonitorKeyboardPower
HP ZR24w HP ZR24w Rosewill RK-9000BR Antec CP-850 
CaseMouseMouse PadAudio
Antec P183 Logitech G700 Razer Goliathus Extended Speed Edition Auzentech HomeTheater HD 
  hide details  
Reply
post #4563 of 7150
I have a few comments about your post:

Quote:
Originally Posted by Kaishi View Post
I have been running my array for about 9 months without a single problem. I have the BBU module, and have both read and write caching enabled. I also have the entire system on a UPS, in a rackmount chassis, mounted in a rack in the basement, which I have leveled to the concrete floor. The whole setup weighs about 300 Lbs. The server is running Windows Server 2008 R2. I'm pleased as punch.
The problem with TLER-type issues is that they will not show any symptoms AT ALL on an array until one of the disks encouters a URE (bad sector, error, read/write delay, however you wish to refer to it). Then when that happens, bang - disk kicked offline, array degraded and vulnerable or worse still failed. This becomes more and more likely as drives age, so you could go for several years before TLER ever becomes an issue for you. Conversely it could be a problem right out of the box.

Quote:
Originally Posted by Kaishi View Post
I acknowledge that if I had a failure, I could lose all my data, and that I would be really sad, possibly miserable. Thing is, it's just data, and life goes on. More importantly, the array is stable, and I have no fear of losing data.
Data might be 'just data', but try explaining that it's just a bunch of 1s and 0s to your parents/partner/children when your array dies and takes with it your graduation/wedding/Toy Story pictures. Nowadays many of our most treasured memories are digital - and losing them (for many people) just isn't worth thinking about.

I think everybody should have at least some fear of losing data. You can think of it like this - a normal person is not afraid of being run over, but it is that small fear of what could happen if you step into the roadway without looking that stops you from being hit by a truck each day. Taking a few small precautions with your data in the knowledge of what could happen is the best way to try to prevent it happening.

Quote:
Originally Posted by Kaishi View Post
TL;DR: I'm running drives that *could* suffer from URE, but I haven't had a problem yet. However, I'm aware of it. The point is to be aware; to know the risks.
Just to note here - ALL drives suffer from UREs, including the most expensive enterprise edition drives. Sooner or later you will encouter an error, the only difference is how well your system can cope with it. With an appreciation of the risks you can design a system to cope with as much as is reasonable to design for for your given situation, and backup whatever else is too important to lose should the nightmare happen and your array fail. For some this means running multiple small RAID6 arrays on RAID edition drives with multiple server and offline backups, for others they may be happy running a single large RAID5 and just backup the odd photo to DVD.
post #4564 of 7150
Thank you BLinux and Kaishi for your input. Seems like RAID 5 at this size is a gamble, it will work but its a gamble and with TLER the odds are not as favorable.

I am now thinking RAID 6 might be the way to go. This would mean I would have to get a Perc 6i. I guess I have to settle with 2TB with RAID 6 instead of 3TB with RAID 5
-Is TLER still an issue with RAID 6?

I really with I could get RE3/RE4 drives.
-I was reading somewhere that either Samsung or Hitachi have consumer grade drives that allow TLER/CCTL/ERC to be enable like the WD Black use to be, any idea which ones?

What other RAID choice I have with 4 drives? I am looking for a RAID that ranks in descending priority: reliability, storage space and then performance.

Thanks
post #4565 of 7150
Quote:
Originally Posted by diehardfan View Post
Thank you BLinux and Kaishi for your input. Seems like RAID 5 at this size is a gamble, it will work but its a gamble and with TLER the odds are not as favorable.

I am now thinking RAID 6 might be the way to go. This would mean I would have to get a Perc 6i. I guess I have to settle with 2TB with RAID 6 instead of 3TB with RAID 5
-Is TLER still an issue with RAID 6?

I really with I could get RE3/RE4 drives.
-I was reading somewhere that either Samsung or Hitachi have consumer grade drives that allow TLER/CCTL/ERC to be enable like the WD Black use to be, any idea which ones?

What other RAID choice I have with 4 drives? I am looking for a RAID that ranks in descending priority: reliability, storage space and then performance.

Thanks
I don't remember right now, but I thought PERC 5 supported RAID-6? I deal with so many PERC 5/6 cards, I don't quite remember...

TLER is an issue with large drives and RAID in general. Error recovery processes take much longer with large drives, hence something needs to be done to "limit" that process so that RAID controller knows the drive is still alive; otherwise it might assume it is dead and mark it as failed. This is not an issue if your drive is small ( less than 250GB? ) or you're not running on a RAID controller. One option, is to use a good software RAID, such as that implemented in Linux or Solaris operating systems, but I assume most people here are concerned with Windows.

The other "reliable" RAID is RAID-10. With 4 drives, RAID-10 may give you better performance than RAID-6 and provide decent redundancy, albeit slightly less. (in RAID-10 w/ 4x HDD, you can survive 2 drives failures as long as they are *not* part of the same mirror, in RAID-6 w/ 4x HDD, *any* two drives can fail and the array can survive.)

Regarding Kaishi's comments: it all depends on how important the data is for the individual or organization that owns the data. I deal with businesses that could stand to lose a lot of money or future income should certain data be lost and not recoverable. people could lose their jobs over this data, so the criticality is probably higher than your average home user. on the other hand, just "being aware" of the URE problem isn't enough. That's a little bit like saying "well, i know i can get in a car accident with serious injury, but I haven't had a car accident in 15 yrs so I'll chance it and not buy auto insurance..." We all know, insurance doesn't seem necessary until you actually need it. So, just because "everything is fine so far" isn't good enough.... when you actually run into a problem, and you lose something that is really is important, your tone will completely change. But, if the data is "just data" and not that important, then it really may not matter.

My general message is this: the seriousness of protecting against data loss is only fully understood when you've actually *had* to deal with it once or twice. a single person's "it hasn't happened to me" is statistically irrelevant, and I would NOT use such statement to make my own decisions.
TAIPEI
(10 items)
 
AURORA
(13 items)
 
 
MotherboardGraphicsRAMHard Drive
ASRock X99 Extreme11 EVGA GTX 980 Superclocked 32GB 8x4GB Corsair LPX Samsung XP941  
Hard DriveCoolingOSMonitor
Western Digital 3TB RE Noctua NH-D15 Fedora 21 Linux Samsung S27D590C 
PowerCase
Seasonic SS-1200XP Cooler Master Cosmos II 
CPUMotherboardGraphicsRAM
Dual Quad-core L5430 2.66Ghz 12mb cache Intel 5000 chipset ATI ES1000 64GB FBDIMM DDR2 PC2-5300 667Mhz 
Hard DriveOSPower
WD3000FYYZ PERC H700 w/ 512MB cache CentOS 7.2.1511 950W x2 
  hide details  
Reply
TAIPEI
(10 items)
 
AURORA
(13 items)
 
 
MotherboardGraphicsRAMHard Drive
ASRock X99 Extreme11 EVGA GTX 980 Superclocked 32GB 8x4GB Corsair LPX Samsung XP941  
Hard DriveCoolingOSMonitor
Western Digital 3TB RE Noctua NH-D15 Fedora 21 Linux Samsung S27D590C 
PowerCase
Seasonic SS-1200XP Cooler Master Cosmos II 
CPUMotherboardGraphicsRAM
Dual Quad-core L5430 2.66Ghz 12mb cache Intel 5000 chipset ATI ES1000 64GB FBDIMM DDR2 PC2-5300 667Mhz 
Hard DriveOSPower
WD3000FYYZ PERC H700 w/ 512MB cache CentOS 7.2.1511 950W x2 
  hide details  
Reply
post #4566 of 7150
I have a perc 5i with 5x1tb hdds in raid 5.

I added another drive last night and the controller is currently rebuilding.

Its 47% through (took 14hrs) and says still 12 hrs to go. Is this normal time how long it takes? I thought it would be 3-4 hrs max since it has its own dedicated parity calc cpu?
post #4567 of 7150
No - that's perfectly normal.

Rebuilds are VERY complex - you are shifting around 5TB worth of data, shuffling it about sector by sector, and that takes time.
post #4568 of 7150
no worries, thanks for the reply.

EDIT: another question, will the rebuild / reconstruct pick up any bad sectors on the discs that might have been quietly sitting there with files that have not been accessed in a while ? Is there a way to check if the SMART report for the attached drives?
Edited by Omarko - 9/16/10 at 2:33pm
post #4569 of 7150
Quote:
Originally Posted by BLinux View Post
I don't remember right now, but I thought PERC 5 supported RAID-6? I deal with so many PERC 5/6 cards, I don't quite remember...
I am basing this on the specs as mentioned on the website
Perc 5i, Perc 6i

PERC 5/i vs 6/i Whitepaper: http://www.dell.com/Downloads/Global...0255-Dixit.pdf

Quote:
TLER is an issue with large drives and RAID in general. Error recovery processes take much longer with large drives, hence something needs to be done to "limit" that process so that RAID controller knows the drive is still alive; otherwise it might assume it is dead and mark it as failed. This is not an issue if your drive is small ( less than 250GB? ) or you're not running on a RAID controller. One option, is to use a good software RAID, such as that implemented in Linux or Solaris operating systems, but I assume most people here are concerned with Windows.
Yes I was recommend ZFS, but I has ZERO experience with Linux or Solaris OS.

Quote:
The other "reliable" RAID is RAID-10. With 4 drives, RAID-10 may give you better performance than RAID-6 and provide decent redundancy, albeit slightly less. (in RAID-10 w/ 4x HDD, you can survive 2 drives failures as long as they are *not* part of the same mirror, in RAID-6 w/ 4x HDD, *any* two drives can fail and the array can survive.)
Think I am going to settle with RAID 6, since reliability is higher priority for me than performance. Is the performance difference b/w RAID 10 and RAID 6 significant, btw? Also, I will be getting 2TB total space with 4 X 1TB with either RAIDs.
post #4570 of 7150
Frys has the Hitachi Deskstar 2TB 3.5" Retail Kit 32MB Buffer, 7200RPM, Hard Drive HD32000IDK7/7K for $104.99, limit 2. But you can always run through a couple times. As of this evening, the frys near my house has "1 available" so I am going to wait until Friday morning to see if they get more in stock for this weekend, and hopefully pull the trigger on 4 more of these suckers.
    
CPUMotherboardGraphicsRAM
Core i7 970 ASUS P6T Deluxe eVGA GTX 680 6x4GB Kingston Hyper X DDR3-1600 
Hard DriveOptical DriveOSMonitor
Boot R0: 2x256GB Corsair SSD; Scratch R0: 3x750GB Samsung 22x LightScribe DVD+/-DL [SATA] Windows 7 x64 Ultimate Dell 2407FPWx2 
KeyboardPowerCaseMouse
Microsoft Sidewinder X6 Antec TruPower Q 1KW Custom Fabricated 18" cube (Mountain Mods UFO) Logitech G9 Laser 
Mouse Pad
Faux Woodgrain (My Desk) 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
Core i7 970 ASUS P6T Deluxe eVGA GTX 680 6x4GB Kingston Hyper X DDR3-1600 
Hard DriveOptical DriveOSMonitor
Boot R0: 2x256GB Corsair SSD; Scratch R0: 3x750GB Samsung 22x LightScribe DVD+/-DL [SATA] Windows 7 x64 Ultimate Dell 2407FPWx2 
KeyboardPowerCaseMouse
Microsoft Sidewinder X6 Antec TruPower Q 1KW Custom Fabricated 18" cube (Mountain Mods UFO) Logitech G9 Laser 
Mouse Pad
Faux Woodgrain (My Desk) 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: RAID Controllers and Software
Overclock.net › Forums › Components › Hard Drives & Storage › RAID Controllers and Software › PERC 5/i RAID Card: Tips and Benchmarks