Seagate 8TB Shingled Drives in UnRAID


Recommended Posts

34 minutes ago, uldise said:

if you are talking about prices, can you compare these Seagates? (these prices are EUR in Latvia/Europe) :

- Archive HDD ST8000AS0002 - 305 EUR

- SkyHawk Surveillance HDD ST8000VX0022 - 305 EUR

- IronWolf ST8000VN0022 - 310 EUR

- Desktop HDD ST8000DM002 - 330 EUR

 

Archive and Desktop are 5900RPM, while Skyhawk and Ironwolf are 7200RPM..

if you wanna low rpm PMR, then Desktop only.. 

 

Tough choice with those prices -- they're essentially all the same price.    If the application is one that the archive drives are appropriate for, that's probably what I'd buy;  otherwise I'd go with the IronWolf.

 

Link to comment
7 hours ago, garycase said:

 

Tough choice with those prices -- they're essentially all the same price.    If the application is one that the archive drives are appropriate for, that's probably what I'd buy;  otherwise I'd go with the IronWolf.

 

Definitely go with the archive for unRAID. For parity maybe try the IronWolf. We have good reliability numbers on the archive disks, but the others generally get crappy scores from Amazon (at least in the States). You might get a bit better write performance from the IronWolf if writing multiple streams to your array simultaneously with the IronWolf, but if that is not so common, I'd probably just go for the archive.

  • Upvote 1
Link to comment
9 hours ago, garycase said:

The agreement WD has with the China's Ministry of Commerce (MOFCOM) required separate assembly lines for 2 years (which expires this year), but did not preclude using the same technologies between the brands.  As I noted above, if you put an 8TD WD Red next to an 8TB HGST UltraStar HE8 they are absolutely identical in every way except for the labeling.   They do, I'm sure, have different firmware and clearly run at different speeds; but I suspect there's little if any difference in reliability -- in fact, the Reds may be the more reliable of the two, simply due to the lower stresses of running at a lower rpm.

 

Indeed, I suspect the new 8TB WD Gold drive (a 7200 rpm helium sealed unit) is very likely identical to the HGST UltraStar -- although it has to be made on a different line since the 2-year hiatus hasn't yet expired.

 

Reading some articles about the WD 8TB's on various IT sites (Storage Review, etc.) you can find the following tidbits r.e. the WD helium sealed units ...

 

"... WD is using their HelioSeal helium-technology to get the higher capacity much like the HGST Ultrastar Helium Drives. ..."

 

" ... The limited information we have ... indicates that it is nearly a carbon copy of the HGST Ultrastar He8 Series  "

 

"  ... WD ... indicated that it is employing technologies across both brands, which includes, but is not limited to, mechanical components, electronics and firmware."

 

 

I very seriously doubt it => I've certainly not read anything to suggest that.   I've installed 7 of these drives -- most in May of last year -- and they're all performing perfectly in 24/7 operation.

 

As I already noted, I'm not saying the archive units aren't excellent drives -- I have a few of them as well; but once the helium-sealed Reds came out, that's all I've been buying, and I'm VERY pleased with their performance.    The cost difference simply doesn't matter to me => but if the extra $75 or so (figure ~ $10/TB) is important to you; then by all means go for the savings.

 

 

A lot of arm waving doesn't make it true. We'll just have to wait for some real reliability stats to come out. Glad yours are working well so far, but a sample size of 7 compared to a sample size of >8000 .... 

 

I did read that the Helium does in fact leak out over time. Maybe around 10 year mark. I have a couple drives in the 8 year range in my backup server. 10 years not out of the question. And what happens as it starts to leak out?

 

But you are paying $100 premium per drive for a slower drive, with no firm reliability stats - times 7 = +$700 for same capacity.


Call me cheap - but I think I could find something better to do with my $700! I could buy 3 more archive drives and still have enough left over for steak at Ruth Chris. :)

Link to comment
1 hour ago, bjp999 said:

But you are paying $100 premium per drive for a slower drive,

 

FWIW the cost difference is more like $75.   And while that is a premium price over the Seagate Archive drive;  it's also $90 LESS than the 8TB HGST UltraStar drive.

So from an apples-apples comparison (8TB Helium-sealed PMR drives), it's $90 CHEAPER than the faster version of essentially the same drive.    Perhaps I should pay the extra $90 to get the HGST version for it's "firm reliability data" (0% failures) :D

Link to comment
7 hours ago, garycase said:

 

FWIW the cost difference is more like $75.   And while that is a premium price over the Seagate Archive drive;  it's also $90 LESS than the 8TB HGST UltraStar drive.

So from an apples-apples comparison (8TB Helium-sealed PMR drives), it's $90 CHEAPER than the faster version of essentially the same drive.    Perhaps I should pay the extra $90 to get the HGST version for it's "firm reliability data" (0% failures) :D

The HGST drives are 7200 RPM, so considerably faster than the REDS and archives. Some premium is justified. But Helium makes me nervous and those drives too pricey for this use. Even BackBlaze only had 45 of them (no where near the 8600 Seagates - it is, after all, a balancing of reliability and cost that drives their buying decisions). But there is another option - HGST makes a non-helium 8T 7200 RPM drive that was recently on sale for 299 (cheaper than the RED). If one needed 7200 RPM, that would seem a good option. (Of course they are unproven with reliability studies, but HGST has earned the benefit of the doubt in my eyes, having released consistently dependable gear (0.60% failure rate as a brand) since the 2T days.) They don't have the "performance wall" and would be my choice if I was spending more.

 

In fact it was my choice as I bought one, currently my standby drive and possible parity2. But realizing that by buying the Seagates I was likely losing 1-2% reliability, but saving 1/3 the cost, seemed a fair exchange. And I bought 2 of those, and looking to complete my disk refresh with a couple more. I would have felt differently if I was giving up a lot more reliability.

 

If I was setting up a new array today, I'd go with the HGST non-helium for parity(ies) and archives for the data drives. And could easily understand people deciding on all archives.

 

Archives can be had exactly $100 less than REDS.

Link to comment
3 minutes ago, tucansam said:

How does the community feel about a WD Red 8TB for parity and the rest of the array being comprised of 8TB Seagate Archives?

I think that'd be a great choice.   In fact, you can even use an archive for parity if you want -- as long as you don't have a lot of simultaneous writes by different processes/users that's very unlikely to hit the shingled performance "wall".

 

As I noted several times above, my personal preference for the helium-sealed units does NOT mean I don't like the archives -- in fact I have several of them, and would still buy more, depending on the projected use of the drive.    Several folks have all-archive servers, and have been very pleased with the performance.

 

Link to comment
On 3/6/2017 at 8:44 AM, tucansam said:

How does the community feel about a WD Red 8TB for parity and the rest of the array being comprised of 8TB Seagate Archives?

 

I'd recommend the HGST 8TB non-Helium for parity.

 

https://www.bhphotovideo.com/c/product/1303685-REG/hgst_0s04012_8tb_3_5_sata_internal.html

 

It is quite a bit faster (7200RPM vs 5400RPM), and would better support simultaneous writes. About the same price as the Red, which is actually slower than the archives. I can see no advantage to using the REDs for parity.

  • Upvote 2
Link to comment
1 hour ago, garycase said:

I think that'd be a great choice.   In fact, you can even use an archive for parity if you want -- as long as you don't have a lot of simultaneous writes by different processes/users that's very unlikely to hit the shingled performance "wall".

 

Actually, the SMR drives like Seagate's 8TB archive, offer better performance than PMR for simultaneous random writes, up to the persistent cache size. The SMR drives put those writes into the persistent cache without the seek overhead of the PMR drives. This is very similar to the performance of MLC/TLC/QLC used as SLC cache on SSDs, for example the Hellfire has ~17GB of cache, less than the Seagate 8TB archive. And WD Black went with ~5.5GB.
 

Edited by c3
  • Upvote 1
Link to comment
16 minutes ago, c3 said:

Actually, the SMR drives like Seagate's 8TB archive, offer better performance than PMR for simultaneous writes, up to the persistent cache size. The SMR drives put those writes into the persistent cache without the seek overhead of the PMR drives. This is very similar to the performance of MLC/TLC/QLC used as SLC cache on SSDs, for example the Hellfire has ~17GB of cache, less than the Seagate 8TB archive. And WD Black went with ~5.5GB.
 

 

True, but if you are copying fresh data to the drive, seems it'd be somewhat unlikely you'd be making hits inside the persistent cache (depends on the block size in the persistent cache compared with unRAID read block size). Drives already have some level of caching which would likely come into play with very localized activity.

 

Has their been any attempt to compare simultaneous write performance on SMR vs PMR parity drives?

Link to comment
15 minutes ago, bjp999 said:

 

True, but if you are copying fresh data to the drive, seems it'd be somewhat unlikely you'd be making hits inside the persistent cache (depends on the block size in the persistent cache compared with unRAID read block size). Drives already have some level of caching which would likely come into play with very localized activity.

The persistent cache is a write cache, there is no "hit" involved.

 

The persistent cache is measured in GB, the cache you are referring to is measured in MB. Might be better to call it disk buffer. The way that buffer is used is different on SMR vs PMR. Using it for writes puts the data at risk to power failures. Persistent cache does not, hence the name persistent.

Link to comment
1 hour ago, c3 said:

The persistent cache is a write cache, there is no "hit" involved.

 

The persistent cache is measured in GB, the cache you are referring to is measured in MB. Might be better to call it disk buffer. The way that buffer is used is different on SMR vs PMR. Using it for writes puts the data at risk to power failures. Persistent cache does not, hence the name persistent.

 

Is the "cache" spinning magnetic media equivalent to PMR media? Or is it SSD-like electronic media?

 

I'm still trying to figure our why a slower rotating drive would be faster for multiple streams.

Link to comment

The persistent cache is an area of the disk that's not shingled -- so yes, it's spinning media.

 

C3's point -- which I hadn't thought of but is very valid -- is that for random writes the shingled drives cache the data in the persistent cache; and then later do the block rewrites needed to move that data to a shingled block.    So if you're doing a lot of random writes, they'll all be to the same area of the disk -- thus eliminating additional seeks that would be needed if the random writes were done directly to the target area.    While a 7200rpm drive clearly writes a bit faster than a 5900rpm drive (e.g. the archive units);  the added seeks would take FAR more time than is saved by the higher rpm.    So indeed, a large number of simultaneous random writes would actually perform better on the shingled archive drives >:(

 

Link to comment
56 minutes ago, garycase said:

The persistent cache is an area of the disk that's not shingled -- so yes, it's spinning media.

 

C3's point -- which I hadn't thought of but is very valid -- is that for random writes the shingled drives cache the data in the persistent cache; and then later do the block rewrites needed to move that data to a shingled block.    So if you're doing a lot of random writes, they'll all be to the same area of the disk -- thus eliminating additional seeks that would be needed if the random writes were done directly to the target area.    While a 7200rpm drive clearly writes a bit faster than a 5900rpm drive (e.g. the archive units);  the added seeks would take FAR more time than is saved by the higher rpm.    So indeed, a large number of simultaneous random writes would actually perform better on the shingled archive drives >:(

 

 

But each write is preceded by a READ from the shingled area. (Only the writes are cached, not the reads). The read will cause the seek to the shingled area, and then the write would cause another seek to the unshingled area where the persistent cache lives. So the farther away from the persistent cache tracks, the slower the unRAID writes would be? The advantage described makes more sense if the persistent cache is electronic, but this actually sounds like it may be a worse penalty than I was thinking.

  • Upvote 1
Link to comment

True if you're using the traditional write mechanism; but many folks these days are using the "reconstruct write" method (aka "Turbo Write") ... and I'd think that those with use cases where there are frequent simultaneous writes from multiple processes/users are even more likely to enable this.

 

With turbo write enabled, the speed advantage outlined by c3 would indeed result in quicker writes.   With the traditional read/modify/write method that advantage is indeed largely gone -- although the "penalty" isn't as bad as it might seem, since much of the delay in seeking to the persistent cache area is using the same time that a PMR drive would be waiting for a rotation of the drive.

 

But I agree the advantage c3 outlines is really only true if the write method is set to reconstruct write.

 

In general, I agree it's better to use a PMR unit for parity if your use case has frequent simultaneous writes.

 

Link to comment
11 hours ago, bjp999 said:

 

I'd recommend the HGST 8TB non-Helium for parity.

 

https://www.bhphotovideo.com/c/product/1303685-REG/hgst_0s04012_8tb_3_5_sata_internal.html

 

It is quite a bit faster (7200RPM vs 5400RPM), and would better support simultaneous writes. About the same price as the Red, which is actually shower than the archives. I can see no advantage to using the REDs for parity.

Have two of those exact drives from B&H.  They looks identical to the two WD Red 8TB non-pro Red drives I got from SuperBiiz except for the label.  So the WD Red non-Pro 8TBs are helium filled?

 

 

Edited by BobPhoenix
Link to comment
1 hour ago, BobPhoenix said:

Have two of those exact drives from B&H.  They looks identical to the two WD Red 8TB non-pro Red drives I got from SuperBiiz except for the label.  So the WD Red non-Pro 8TBs are helium filled?

 

 

 

Is the point here that the Helium filled RED and the non-Helium filled HGST look the same? And therefore you can't tell by looking at them what's on the inside?

Link to comment
20 hours ago, bjp999 said:

 

Is the point here that the Helium filled RED and the non-Helium filled HGST look the same? And therefore you can't tell by looking at them what's on the inside?

Not exactly.  I was wondering if only the WD Red Pro versions were the ones to be Helium filled and the non-pro Reds were like the non-helium HGST.

Link to comment
10 minutes ago, BobPhoenix said:

I was wondering if only the WD Red Pro versions were the ones to be Helium filled and the non-pro Reds were like the non-helium HGST.

 

Both are helium-sealed units => the difference is the non-pro run at 5400 rpm vs. 7200 rpm for the Pro version; and the warranty is 2 years longer for the Pro version (5 yrs vs. 3).

 

Link to comment

Pardon me for asking a few questions which might already be covered in the 13 pages of this thread, but here goes anyways (while I go back to the first page and start re-reading the entire thread).

 

My current array is 5 drives of 4TB HGST 7200rpm that have parity check times around 8 hr, 43 min, 42 sec. My 4 drives are a little over 2 years old and looking perfect, however I'm running out of space. My options are to add another 4TB HGST drive or to expand to larger drive sizes. All the talk of larger drives and solid performance have me considering a large size upgrade. The 8TB Seagate Archive drives are so dirt cheap (~ $225 after 10% off coupon) compared to other options, so they're very tempting, but I have a question about how it might impact parity checks. My last upgrade was swapping in 4TB 7200rpm for old 2TB 5900RPM where despite doubling the size of the data array the parity checks remained nearly the same time.

 

What are the parity check times of arrays that only run 8TB Seagate Archive drives? How does that compare to non-archival 8TB arrays?

 

 

Link to comment
10 minutes ago, BRiT said:

What are the parity check times of arrays that only run 8TB Seagate Archive drives? How does that compare to non-archival 8TB arrays?

 

The ones I've seen reported were a bit over 15 hours. on arrays with all 8TB Seagate archive drives.

 

  • Upvote 1
Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.