Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

We're acquiring several 1U servers with (8) 2.5" drive slots. Although we can use either SATA or SAS, there is a large price variance as soon as you order 16 or 24 of these drives, so we are looking at the 2.5" SATA interfaced drives.

I know that Seagate and WD both make "Enterprise" 2.5" drives, which are fast (10k and 15k RPM), but are also fairly expensive.

What issues would we run into using 7200RPM 2.5" non-Enterprise drives? By the way, these will be hooked up to a RAID controller (though, they may just be configured as JBOD). These drives are almost $100 lower in price, per drive.

share|improve this question
7  
It's like the difference between putting golf cart tires on your car. Sure they'll work, but I wouldn't want to ride in it. –  joeqwerty Aug 16 '11 at 15:48
add comment

6 Answers 6

up vote 14 down vote accepted

In addition to the problems above, you may have additional issue running these drives in RAID configuration due to the lack of TLER. (If you are considering a model without.) This quote references desktops and the RAID Edition drives but I imagine the same to be true in the 2.5" line if you substitute in "notebook" and "enterprise" or "SAS" where applicable.

Western Digital manufactures desktop edition hard drives and RAID Edition hard drives. Each type of hard drive is designed to work specifically in either a desktop computer environment or a demanding enterprise environment.

If you install and use a desktop edition hard drive connected to a RAID controller, the drive may not work correctly unless jointly qualified by an enterprise OEM. This is caused by the normal error recovery procedure that a desktop edition hard drive uses.

When an error is found on a desktop edition hard drive, the drive will enter into a deep recovery cycle to attempt to repair the error, recover the data from the problematic area, and then reallocate a dedicated area to replace the problematic area. This process can take up to 2 minutes depending on the severity of the issue. Most RAID controllers allow a very short amount of time for a hard drive to recover from an error. If a hard drive takes too long to complete this process, the drive will be dropped from the RAID array. Most RAID controllers allow from 7 to 15 seconds for error recovery before dropping a hard drive from an array. Western Digital does not recommend installing desktop edition hard drives in an enterprise environment (on a RAID controller).

Western Digital RAID edition hard drives have a feature called TLER (Time Limited Error Recovery) which stops the hard drive from entering into a deep recovery cycle. The hard drive will only spend 7 seconds to attempt to recover. This means that the hard drive will not be dropped from a RAID array. Though TLER is designed for RAID environments, it is fully compatible and will not be detrimental when used in non-RAID environments.

share|improve this answer
    
Very insightful ! –  Antoine Benkemoun Aug 16 '11 at 16:21
    
Great info. Thanks! –  Matt Simmons Aug 16 '11 at 17:13
    
That's the only quantitative data that shows a real difference between desktop and enterprise models (apart from warranty time and price) –  Hubert Kario Aug 16 '11 at 17:32
    
I wouldn't say that necessarily. There's a huge difference in performance that is easily measurable. –  Aaron Copley Aug 16 '11 at 18:06
    
My main concern was that it was a limited command set in the drives. Of course, if we're doing software RAID, I'm not sure how much that will matter. –  Anthony Aug 16 '11 at 18:49
show 2 more comments

The biggest difference? Failure-rate.

Those 'enterprise' drives are warrantied for 5 years, whereas the cheaper ones are probably warrantied for less. Also look into the spec-sheets for them and look at their duty-cycles. The Enterprise drives are designed to run for 5 years straight, where the 'desktop' drives are designed to run 8 hours a day for 5 years. Very different use-cases and will impact your drive failure rates.


A second thing to look at is a line on those spec-sheets named "Nonrecoverable Read Error rate", which is a measure of the frequency of bits that are unable to be read inside the recovery window.

As of this posting (8/16/2011), the Seagate Savvio 10K.5, a 10K RPM Enterprise 2.5" drive, has its rate listed as 1x10^16. The Western Digital Scorpio Black, a 7.2K RPM consumer oriented 2.5" drive, has its rate listed as 1x10^14 bits. By this measure, the Savvio drive is two orders of magnitude more reliable.

This error rate puts an upper limit on how large of a RAID5 set you can build with such drives. When a drive fails in a RAID5 array, the array then has to read the entire RAID volume in order to rebuild the parity. If a non-recoverable read error occurs, you can lose the entire RAID set. Some RAID cards can get around this, others can't. They're not all built the same.

The above error-rate measures are approximate, but are the point where such errors are more likely to happen than not.

  • 10^14 bits = 12.5 TB
  • 10^16 bits = 1.25 PB

Only, you don't want to build arrays that large. The largest you want to build them is about 50% that size to minimize the likelihood of the rebuild failing. For those really cheap 1TB 2.5" drives, you can only fit 7 of those in a R5 array, where with the more expensive 10K RPM drives you could fit 15 of those 900GB drives in an array and feel safe in the knowledge that it'd rebuild just fine (but take a long time); your parity-losses are worse with the cheaper drives, which impacts your overall capacity.

share|improve this answer
    
there were few studies that showed no difference in reliability between enterprise and user models. The Nonrecoverable Read Error rate is different but mostly depends on model, not enterprise/desktop. What is important is the amount of time a desktop drive and enterprise one will take to try read data from damaged sector. You don't want your RAID array stalled for 4 or 5 seconds while one of the drives tries to read data... –  Hubert Kario Aug 16 '11 at 17:30
    
@Hubert Kario, how about a link to these studies? –  Zoredache Aug 16 '11 at 18:42
    
Any links to any of this data? We've run three servers in RAID-10 (4 drives each), using software RAID, and consumer drives, for almost 4 years now - the only failure we've had is from Seagate drives. WD consumer drives have kept ticking away. –  Anthony Aug 16 '11 at 18:51
    
cs.cmu.edu/~bianca/fast07.pdf "Disk failures in the real world: What does an MTTF of 1,000,000 hours mean to you?", it's in the bibliography of the Google disk study "Failure Trends in a Large Disk Drive Population" labs.google.com/papers/disk_failures.pdf , also a good read –  Hubert Kario Aug 17 '11 at 6:09
add comment

This is a really bad idea. There's a good reason laptop drives are so much cheaper than their server-grade cousins. Plain and simple - they are not built to be used 24x7. You will see incredibly high failure rates with these drives if used in a server capacity.

share|improve this answer
    
10k and 15k are different, thats for sure, the difference between 7.2k enterprise and desktop models is at best a different firmware. –  Hubert Kario Aug 16 '11 at 20:44
add comment

One thing to take into account regarding the failure rate on drives designed for RAID vs single drive systems is the vibration factor. Raid drives are designed to handle the additional vibration caused by a cage full of drives where as consumer class drives are not.

share|improve this answer
add comment

It depends entirely on what you're trying to do with the server.

If it's a single node in a large failure resilient cluster (like hadoop or some such), that's fine.

If the drives are just for booting and swap, and all the real data is on a nice reliable SAN or enterprise NAS system, again, it should be just fine (but in that case you only need two or three drives, so why bother skimping?)

If you're just trying to be cheap because your budget is tight, make sure you have a hot spare (or three) in the raid array, and know that this could mean a frantic 2AM drive to work if two drives two fails at once.

The enterprise drives are preferable, and there for a reason, but if you're honest with yourself about the reasons you're not using them, and PLAN for the much more likely failure rate, go ahead.

share|improve this answer
add comment

IOPS - how many do you need? Throughput - how much do you need?

If you're doing a refresh, and you're currently using ATA drives in your existing boxes, is your storage a constraining performance factor? If not, you could probably stick with the ATA drives, though as others point out, you may see a slightly higher failure rate.

If you're not sure about performance issues, look at perfmon if it's a Windows physical server, or whatever the equivalent would be for Linux. Virtual server products have their own utils to see disk performance. You can google for the metrics you'll want to investigate.

SAS has better IOPS & throughput, resulting from faster seeks & spindle speeds. Also, "enterprise" products will likely have better algorithms for dealing with cache (when to go ahead and write, when to accumulate writes before comitting, etc.).

But again, if storage performance isn't a factor, then don't spend money upgrading it. Or even it if is, you may get what you need by just using more slower cheap spindles than fewer faster, but more expensive spindles. Find out what your performance bottlenecks are, and spend money there.

share|improve this answer
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.