Which WD drive for parity? Black or Purple?

[fitbrit]

I figured that a purple drive might be good for parity, since any write to the unraid server is also causes a write to parity too. So, it should be the most active of disks int he system overall. I currently have a 4TB black as parity, but am thinking of upgrading to 6 TB (or more?). Any thoughts?

[garycase]

I'd stay with the Black series => faster and a better warranty.

 

[garycase]

Actually, WD doesn't offer a 6TB Black (yet).    I assumed from your question that there was a choice ... but that's not the case.

 

In that case, I'd use a 6TB WD Red or the faster (7200rpm) 6TB HGST NAS unit.

[Or wait for a 6TB Black, which I suspect won't be too far in the future  ]

 

 

 

[fitbrit]

Thanks. I'll take a look at the HGST NAS one then.

[garycase]

If you're in the US, you can get it here:

http://www.newegg.com/Product/Product.aspx?Item=N82E16822145973

[Brucey7]

The Purple drive is not ideal.  It's cache is optimised for sequential reads and writes.  It's not designed for random I/O.

[ysss]

Another downside to black is that they run hotter due to 7200rpm.... but then again, according to Back Blaze's data, drive temperature has little bearing to their longevity at all. Presumably as long as it stays within the operating range, and that operating range is wiiiiiiiiide.

[fitbrit]

I'm in Canada, and got the HGST from Newegg.ca. Thanks for the tip and comments Gary and everyone else.

[johnny121b]

I recall reading that the PURPLE drive series, being intended for video recording, has very weak error-checking firmware.  Theory being- that for video recording, it's more important that the drive be able to run 24/7, while dropping the occasional bit was okay.

[WeeboTech]

Thanks. I'll take a look at the HGST NAS one then.

 

I'm pleased with my HGST NAS 6TB 7200 RPM Drives as parity!

up to 225MB/s on the outer tracks!

[garycase]

Thanks. I'll take a look at the HGST NAS one then.

 

I'm pleased with my HGST NAS 6TB 7200 RPM Drives as parity!

up to 225MB/s on the outer tracks!

 

Very nice performance.    A system with all HGST 6TB drives would likely have some excellent parity check and rebuild times.    It'd be interesting to see the write speeds with those as well ... I suspect the typical 35-40MB/s may jump up ~ 50MB/s thanks to their faster seek times and lower latencies.

 

[WeeboTech]

Thanks. I'll take a look at the HGST NAS one then.

 

I'm pleased with my HGST NAS 6TB 7200 RPM Drives as parity!

up to 225MB/s on the outer tracks!

 

Very nice performance.    A system with all HGST 6TB drives would likely have some excellent parity check and rebuild times.    It'd be interesting to see the write speeds with those as well ... I suspect the typical 35-40MB/s may jump up ~ 50MB/s thanks to their faster seek times and lower latencies.

 

I burst write up to 1GB at up to 120MB/s, After that it drops to around 80, then 60, Usually hovers around 40-60 for long running transfers.

[garycase]

Those are certainly VERY good write speeds to a protected array.    Is this using the reconstruct-write mode?

 

[RobJ]

I recall reading that the PURPLE drive series, being intended for video recording, has very weak error-checking firmware.  Theory being- that for video recording, it's more important that the drive be able to run 24/7, while dropping the occasional bit was okay.

 

Can this information be confirmed?  I'm wondering if we should be warning users against this drive, marking it as NOT recommended in the HCL page.

[garycase]

I recall reading that the PURPLE drive series, being intended for video recording, has very weak error-checking firmware.  Theory being- that for video recording, it's more important that the drive be able to run 24/7, while dropping the occasional bit was okay.

 

Can this information be confirmed?  I'm wondering if we should be warning users against this drive, marking it as NOT recommended in the HCL page.

 

I don't think that's really an issue.  Drives that are designed for surveillance tend to have their cache algorithms optimized for video frame storage, so any recalibration activity will not result in lost frames.    I don't think they compromise on error correction for the writes.    I think these are simply the next generation of the earlier WD AV drives that were designed for audio/video applications (including surveillance).

 

Clearly the BEST choice for UnRAID systems are the NAS drives, which are optimized for NAS applications like UnRAID ... but a lot of folks prefer to use the lower cost desktop drives, and haven't had major issues with data loss from these.    I think the Purple's will work just as well (if not better) than the desktop class units.

 

[c3]

 

Clearly the BEST choice for UnRAID systems are the NAS drives, which are optimized for NAS applications like UnRAID ... but a lot of folks prefer to use the lower cost desktop drives, and haven't had major issues with data loss from these.    I think the Purple's will work just as well (if not better) than the desktop class units.

 

I'd like to see the data supporting this claim on NAS drives in unRAID.

 

In a NAS, all the drives are work together to deliver data. This is unlike unRAID were a single drive is used for streaming media. Conversely, two drives are used to write data, unlike NAS were all drives are used. Additional, unRAID drives are subjected to many more spinup cycles vs NAS (or video). These differences certainly move the unRAID workload from typical NAS towards Desktop.

[garycase]

 

Clearly the BEST choice for UnRAID systems are the NAS drives, which are optimized for NAS applications like UnRAID ... but a lot of folks prefer to use the lower cost desktop drives, and haven't had major issues with data loss from these.    I think the Purple's will work just as well (if not better) than the desktop class units.

 

I'd like to see the data supporting this claim on NAS drives in unRAID.

 

In a NAS, all the drives are work together to deliver data. This is unlike unRAID were a single drive is used for streaming media. Conversely, two drives are used to write data, unlike NAS were all drives are used. Additional, unRAID drives are subjected to many more spinup cycles vs NAS (or video). These differences certainly move the unRAID workload from typical NAS towards Desktop.

 

Many NAS units are designed to spindown when not being accessed, so I'm not sure that's as big a difference as it may seem.    Depends on the specific units involved and how they're configured (including, of course UnRAID, as it can also be configured for various spindown times ... or even to never spindown).

 

As for whether the differences move the balance towards desktop drive characteristics ... I don't think so.    Using WD's drive as an example, the Reds (NAS drives) are spec'd for TWICE the number of load/unload cycles as their desktop units (600,000 cycles vs. 300,000 for both Greens and Blacks), have a higher MTBF, and have an operating temperature range that's much better than their desktop cousins [0-65C  vs. 0-60C for the Greens and 5-55C for the Blacks].

 

[c3]

 

Clearly the BEST choice for UnRAID systems are the NAS drives, which are optimized for NAS applications like UnRAID ... but a lot of folks prefer to use the lower cost desktop drives, and haven't had major issues with data loss from these.    I think the Purple's will work just as well (if not better) than the desktop class units.

 

I'd like to see the data supporting this claim on NAS drives in unRAID.

 

In a NAS, all the drives are work together to deliver data. This is unlike unRAID were a single drive is used for streaming media. Conversely, two drives are used to write data, unlike NAS were all drives are used. Additional, unRAID drives are subjected to many more spinup cycles vs NAS (or video). These differences certainly move the unRAID workload from typical NAS towards Desktop.

 

Many NAS units are designed to spindown when not being accessed, so I'm not sure that's as big a difference as it may seem.    Depends on the specific units involved and how they're configured (including, of course UnRAID, as it can also be configured for various spindown times ... or even to never spindown).

 

As for whether the differences move the balance towards desktop drive characteristics ... I don't think so.    Using WD's drive as an example, the Reds (NAS drives) are spec'd for TWICE the number of load/unload cycles as their desktop units (600,000 cycles vs. 300,000 for both Greens and Blacks), have a higher MTBF, and have an operating temperature range that's much better than their desktop cousins [0-65C  vs. 0-60C for the Greens and 5-55C for the Blacks].

 

I'd like to see the data supporting this claim on NAS drives in unRAID.

 

[garycase]

Clearly the manufacturer's specs aren't always supported by real life experiences ... but they're about all we have to go on.    But I don't think the major manufacturers "fudge" their specs to favor one series over another -- i.e. I don't think WD improved the stated specs of the Red series just as a marketing exercise 

[johnny121b]

But I don't think the major manufacturers "fudge" their specs to favor one series over another -- i.e. I don't think WD improved the stated specs of the Red series just as a marketing exercise 

 

That's an awfully gracious stance- given how companies tend to regard customer/product support after-the-sale.  I'm not trying to impune WD.  Quite the contrary, I've had good luck with warranty issues with them in the past.  You have to admit- slashing warranties by 80% and giving almost no real data to differentiate each series....doesn't promote warm, fuzzy feelings.  I regard hard drive marketing akin to wireless router marketing....which is to say- you can't believe a word of it.  A nice graphic (or a [glow=red,2,300]color[/glow]) doesn't equal "better product"

[garycase]

But I don't think the major manufacturers "fudge" their specs to favor one series over another -- i.e. I don't think WD improved the stated specs of the Red series just as a marketing exercise 

 

That's an awfully gracious stance- given how companies tend to regard customer/product support after-the-sale.  I'm not trying to impune WD.  Quite the contrary, I've had good luck with warranty issues with them in the past.  You have to admit- slashing warranties by 80% and giving almost no real data to differentiate each series....doesn't promote warm, fuzzy feelings.  I regard hard drive marketing akin to wireless router marketing....which is to say- you can't believe a word of it.  A nice graphic (or a [glow=red,2,300]color[/glow]) doesn't equal "better product"

 

I agree in principle ... I don't think the marketing "hype" of the drives is all real by any means.  But I DO think there's an actual difference between the various models -- not just a different color "sticker" on the drive ... and that the specifications are based on their engineering analyses of the designs.  The NAS units almost certainly have an improved actuator design that accounts for the much higher (double) load cycle specifications.    And they DO "put their money where their mouth is" with regards to warranting these drives longer than their Green desktop cousins.

 

... and one thing I'll definitely give WD credit for => they have superb warranty policies.  A painless online RMA process -- and very quick replacements (within a couple days if you use the Advance RMA process).

 

[Walter S]

For what its worth I'm using a WD Enterprise 3TB for parity.  I think there supposed to be as bullet proof as you can get for a server..

its been good for me..

[garycase]

For what its worth I'm using a WD Enterprise 3TB for parity.  I think there supposed to be as bullet proof as you can get for a server..

its been good for me..

 

Agree the enterprise class drives are superb ... there's a reason WD warrants them for 5 years.  These drives also have error rates a full order of magnitude better than the desktop and NAS class drives (<1 in 10^15  instead of <1 in 10^14), have the longest MTBF of any WD drives, and are spec'd for the same 600,000 load cycles as the NAS units.