High performance Unraid server workbook

[mjstumpf]

Regarding SSD-as-cache:

 

One should be quite careful here--I have friends that deal with mass distribution of these, and they tell me to be quite cautious as the failure rate in most brands is unbelievably high.  (I'm asking for more details; Kingston and Samsung were the only exceptions; but some Kingston lines appear to be rebadged Intel, so I wonder if all of kingston is secure)  

 

Now, knowing how they work, I would suspect that a teracopy-style-write would fail in such a failure mode, which is what you want.  I also wonder if my style of usage (occasional, large file storage, NOT as a high-write, sparse SQL db drive) would also mean I could pick anything and it would last forever.

 

But then there's the cost.  $130 for a 60gb (on sale) + $130 for another 60gb + $20 for a 2-port SATA and we're talking about another $300 for _convenience_; I'm not sure that this is worthwhile, if I've driven other components (because it's a new array purchase) toward 40mb/sec sustained write performance.  Frankly, for $300, I'd almost rather have another 4gb of storage.  I'm more inclined to stick spare 500gb drives in as cache.

 

 

 

 

[WeeboTech]

writing to a raid1 is slower than writing to a single drive.

 

Are you sure about that?  What are you basing that on?  I actually don't believe that's the case (in any significant amount.  If you tell me it dings 1 MB/sec in overhead or adds some small level of throughtputvariance, ok).  When writing blocks out, there is no dependency between the drives from a driver's perspective.  

 

 

There is a dependency, but it's hardly noticeable these days, plus you usually pick matching drives, so it's rarely an issue.

I do sw raid1 with scsi, sas, sata, ide drives. It's always been a slight performance penalty, but nothing to ever be an issue. Even with vmware volumes on the inside tracks.

 

 

 

Ability to soft raid-1 the cache data would be ideal, yeah.  Kills another sata port, but I'd probably still do it.

 

In the days to come when if we get a unionfs system in unraid, this will be needed to safe guard some data on cache.

 

I was more worried about corner (inflection point) cases inside the mover script:  Start, but don't finish transfering a file, mover "moves" incomplete file, file xfer finishes, mover declares victory, deletes complete file.  But, having looked at the mover script, I don't think that can happen.  I'm going to test it a few times with big files though ;-)

 

There are some safeguards in the mover script to prevent moving open files.

 

 

[WeeboTech]

Regarding SSD-as-cache:

for $300, I'd almost rather have another 4gb of storage.  I'm more inclined to stick spare 500gb drives in as cache.

 

High amounts of ram does not seem to help with transfer speeds. I think the cache gets flushed very often which tends to interfere with transfer speeds (at least from some of my tests).

 

From what I've been reading the samsung spinpoint F3 drives are amoung some of the highest benchmarked drives for write throughput these days.  can't even find the 500 gb drives. I bought a pair of 1tb which use the 500gb platters.

[mjstumpf]

Regarding SSD-as-cache:

for $300, I'd almost rather have another 4gb of storage.  I'm more inclined to stick spare 500gb drives in as cache.

 

High amounts of ram does not seem to help with transfer speeds. I think the cache gets flushed very often which tends to interfere with transfer speeds (at least from some of my tests).

 

From what I've been reading the samsung spinpoint F3 drives are amoung some of the highest benchmarked drives for write throughput these days.  can't even find the 500 gb drives. I bought a pair of 1tb which use the 500gb platters.

 

Ack.  I meant 4TB of drive storage.

 

You know, it's funny.  I've read on this forum that the Samsung F3 were among the worst performing.  But here's the thing:  A high level view of newegg comments seems to indicate that they don't suffer the obscene failure rate afflicting other 2tb drives, so I'm likely buying only them.

 

 

[WeeboTech]

You know, it's funny.  I've read on this forum that the Samsung F3 were among the worst performing.  But here's the thing:  A high level view of newegg comments seems to indicate that they don't suffer the obscene failure rate afflicting other 2tb drives, so I'm likely buying only them.

 

I had the 7200rpm 1tb drives in my mind. Not the ecogreen drives.

Everywhere I've read so far said they are top performers.

 

[bubbaQ]

What this community seems to be missing (probably because the high speed is new/recent) is a "recipe" for constructing the highest-possible-speed-Unraid.

 

No secret.... use 4.5.3 or later and just populate your system with these:  http://www.newegg.com/Product/Product.aspx?Item=N82E16820233112

[mjstumpf]

Update on this:  got the WD EARS drives and the Samsung Spinpoint.  Windows over-the-wire writes go 40mb/sec typical, and iozone posts the spinpoint as slightly faster (36 mb/sec write) than the WD.

 

To me, that makes the Spinpoint an obvious choice; it appears to have a very low failure rate.  It's even on sale at newegg right now for $150. 

[bcbgboy13]

To me, that makes the Spinpoint an obvious choice; it appears to have a very low failure rate.  It's even on sale at newegg right now for $150.   

 

Keep in mind two things - how many people are buying WD and how many are buying Samsung (the same failure rate will be reflected in more negative posts for the higher volume sold)

And I personally do not like the way Newegg ships multiple OEM HDs - I think there is a higher possibility for damage from their way of packing these.

Also when you buy a single drive you often may not go for the lowest priced one as there will be other considerations - speed, warranty, etc.

But when you buy multiple units especially for our purpose the price you pay may take a priority and when you order the WDs they may take a hit during the shipping due to the Newegg's way of packing but of course you will blame the WD reliability...

 

[mjstumpf]

My comments were not made as idle speculation.  I tried to beat the odds 6 months ago, and I failed rather spectacularly.  Newegg shipped me one drive that lasted an hour before click-of-deathing, and the second drive lasted a week.  That's 2/2 DOA, straight from the factory.  If you read the comments carefully you can see enough reports to realize there are grievous problems.  Don't believe me?  Call me crazy and try your luck.  Failure mode tends to be slowly-worsening performance followed by "endless click of death".  Good luck.

 

Newegg actually improved their packaging a bit.

 

As for whether or not samsung drives are indeed higher quality, we'll see.  I've ordered a bunch of them.  I've never seen such a clean Newegg comments set.

 

 

[terrastrife]

unless your falure rate sample size is over 1 million drives personally. you dont count. for example ive gone though over 50 dds teh last year or so, and ive had mainly x brand failures. does this mean x is better htan y? no, it means that ive had mroe of x fail, thats all. there is no inherrent higher failure from x or y, thats just what fate has given me.

[mjstumpf]

unless your falure rate sample size is over 1 million drives personally. you dont count. for example ive gone though over 50 dds teh last year or so, and ive had mainly x brand failures. does this mean x is better htan y? no, it means that ive had mroe of x fail, thats all. there is no inherrent higher failure from x or y, thats just what fate has given me.

 

You know, I've ignored a lot of this kind of stuff so far, but you just need to shut the hell up.  I included plenty of qualification in my comments:  If you think I'm wrong, fine.  Go buy whatever you like.  Why isn't that good enough for you?  Why must you swing your cock around with ridiculous statements like these?

 

You can't be bothered to read, carefully, what I said and why I said it.

 

Oh, and while of course, "I don't count", so far badblocks sweeps have turned up errors in both WD drives, but none in the Samsung.  Take that for what it's worth.

 

 

[terrastrife]

unless your falure rate sample size is over 1 million drives personally. you dont count. for example ive gone though over 50 dds teh last year or so, and ive had mainly x brand failures. does this mean x is better htan y? no, it means that ive had mroe of x fail, thats all. there is no inherrent higher failure from x or y, thats just what fate has given me.

 

You know, I've ignored a lot of this kind of stuff so far, but you just need to shut the hell up.  I included plenty of qualification in my comments:  If you think I'm wrong, fine.  Go buy whatever you like.  Why isn't that good enough for you?  Why must you swing your cock around with ridiculous statements like these?

 

You can't be bothered to read, carefully, what I said and why I said it.

 

Oh, and while of course, "I don't count", so far badblocks sweeps have turned up errors in both WD drives, but none in the Samsung.  Take that for what it's worth.

 

 

 

but thats what, 2 drives? 4 drives? 50 drives? with such a small sample size your stats mean nothing. if im going to quote like you then for me i have 100% samsung failure rate, about 5% wd and 0% hitachi, seagate rates. now, this makes samsunglook horrible, but ive onyl recently purcahsed ONE drive and it died pretty quick, so, 100%. you may buy another 100 wd drives and not have another failure, that would change what you would say.

 

without a very large sample size stats dont mean much.

[Kaygee]

The problem with newegg reviews etc... is that items like HDDs that are commodity items usually only get bad reviews because an item was faulty. The exceptions being exceptional units where people say wow this drive really is fast or wow really low power, wow really quiet.

 

Statistically it would be interesting to see how many units newegg has sold vs how many complaints. If it has sold 5000 hdds and 50 people have complained then you could quantify the results.

 

I have a lot of sympathy with mjstumpf. If i'd bought two drives and both failed I'd be posting also, however this could just be bad luck as terrastrife has pointed out.

 

However expecting one person to have tested 1 million drives is rather ridiculous, AFAIK Michael Dell isn't on this forum, telling someone they dont count isnt going to improve their mood.

 

Seagate and WD have both recently been making wrong decisions. As the the Quantum and Maxtor corporations proved proved making wrong decisions in the HDD market means the sale of your business. 

[WeeboTech]

Seagate and WD have both recently been making wrong decisions. As the the Quantum and Maxtor corporations proved proved making wrong decisions in the HDD market means the sale of your business. 

 

Huh   please elaborate.

[mjstumpf]

The problem with newegg reviews etc... is that items like HDDs that are commodity items usually only get bad reviews because an item was faulty. The exceptions being exceptional units where people say wow this drive really is fast or wow really low power, wow really quiet.

 

Statistically it would be interesting to see how many units newegg has sold vs how many complaints. If it has sold 5000 hdds and 50 people have complained then you could quantify the results.

 

I'm not so sure about this; Newegg, and to a lesser extent Amazon seem to have vibrant, active communities that, I suspect because they're so established, actually prompt people to review--good and bad.  That said, you're not wrong.  It's a hard-to-know problem.  But it's not unknowable.

 

Drive failures tend to occur in batches due to manufacturing defects.  That's why it's good advice to buy drives over time.  It's actually also good advice to buy older, established drive models that have had the kinks worked out.  At this moment, I'm trying to not follow either of these.

 

By reading reviews, it is very clear that _at the very least_ lots of bad WD, and especially seagate LP 2tb drives are being shipped.  When you have people buying 8, and 6 of them fail within a month, and this is not an isolated review (UPS kickball match involving that pkg), there is not just a problem--there is a HUGE manufacturing/QA problem.  Maybe that problem is the electronics being hypersensitive to power supply (as one Seagate review suspected; they had 4 "bad drives", swapped PS to a newer, beefier, one; problem GONE); maybe the problem is newegg "packaging".  Maybe the problem is, that because these drives take so damned long to do surface scans, as of 6 months or so ago WD/Seagate stopped bothering to QA a drive before they left the factory, figuring "Screw the customer, we're in a price war".  It would be interesting to see what the SMART data says on delivery.

 

The samsung drives have one review (if memory serves) complaining of failure.  Is that because the drive is more expensive and thus doesn't sell as much?  Yeah, certainly a possibility.  It's still got a large pile of reviews, though.  I think it is a reasonable conclusion to say that my odds are better with the drive that doesn't have many, many reviews indicating 75% failure rate in a month on a sample set size of 8.  I'll take my chances with Samsung for now.

 

This is all academic, anyway.  I may end up with high failure rates from the Samsung, too.  

 

 

[BRiT]

Seagate and WD have both recently been making wrong decisions. As the the Quantum and Maxtor corporations proved proved making wrong decisions in the HDD market means the sale of your business. 

 

What? Can you please provide details?

[Kaygee]

Seagate

 

SD1a firmware. 1.5TB drive stutter.

 

WD

 

Intellispeed, 4K  (Advance file format my arse), load/unload cycle count smart 193 (parked heads), error recovery in raid arrays causing drives to be dropped (green drives in RAID arrays).

[BRiT]

Intellispeed - marketing byproduct that never made it to the product, so you got this one correct.

 

 

To dispute the rest of your incorrect FUD statements:

 

4K sectors is an advance disk format - this is a fact. All SSD use this, and all drive mfgs are moving to it.

 

load/unload cycle count smart 193 -- not an issue at all on EADS drive which is old by today's standard. If there was an issue in the older drives (EACs ?), they have since corrected it.

 

error recovery in raid array - This is a byproduct of users buying the wrong drive for the wrong purpose. This is exactly why WD markets the "RE" Raid Edition drives. All drives by all mfgs will encounter this issue if they take even 1 ms longer than expected to respond when encountering a real issue.

[terrastrife]

SD firmware is from 2008.