Personal computing discussed

Moderators: renee, morphine, Steel

  • 1
  • 2
  • 3
  • 4
  • 5
  • 7
 
Chrispy_
Maximum Gerbil
Posts: 4667
Joined: Fri Apr 09, 2004 3:49 pm
Location: Europe, most frequently London.

Re: Blackblaze's 2018 HDD stats

Wed Mar 27, 2019 10:35 am

The problem is that increased RPM doesn't translate linearly to more performance. Higher RPM drives typically have lower areal density so whilst the lifespan/power/noise increases are linear (or even exponentially worse) with increased RPM, the performance gains are less than linear.

In theory a 7200rpm drive is 33% faster than a 5400rpm drive but in practice the real advantage is often 15% or less.
Congratulations, you've noticed that this year's signature is based on outdated internet memes; CLICK HERE NOW to experience this unforgettable phenomenon. This sentence is just filler and as irrelevant as my signature.
 
meerkt
Graphmaster Gerbil
Topic Author
Posts: 1487
Joined: Sun Aug 25, 2013 2:55 am

Re: Blackblaze's 2018 HDD stats

Wed Mar 27, 2019 12:04 pm

I never noticed a trend to not have higher densities in 7200 rpm? I think they may sometimes start by introducing new higher density platters in lower rpm drives, but 7200 drives always follow.

The data on https://rml527.blogspot.com/ supports that impression.

BTW, apparently there are not only 5900 rpm drives, but also 5000.
 
The Egg
Gold subscriber
Minister of Gerbil Affairs
Posts: 2937
Joined: Sun Apr 06, 2008 4:46 pm

Re: Blackblaze's 2018 HDD stats

Wed Mar 27, 2019 12:42 pm

I believe they also made 2.5” drives in 4200rpm for a time.

In theory, a lower rpm drive would have less wear and tear, but 7200rpm models are usually marketed as a higher-end product (often with longer warranties), so the drives aren’t necessarily being equipped with the same quality of hardware.
 
Topinio
Gerbil Jedi
Posts: 1758
Joined: Mon Jan 12, 2015 9:28 am
Location: London

Re: Blackblaze's 2018 HDD stats

Wed Mar 27, 2019 5:30 pm

Slower drives are usually consumer junk with much lower quality internals, partly for engineering and partly for value engineering reasons.

Sure, there's less stress so it doesn't need to be so well engineered; this just means that the wear on the cheapo internals from running at 5400 for a couple of years can be expected to result in similar mortality rates to the wear and tear deaths of the much pricier drives from running at their 15000 RPM for much longer periods... see the warranty length for the beancounters' first order stab at it...
Desktop: E3-1270 v5, X11SAT-F, 32GB, RX 580, 500GB Crucial P1, 250GB Crucial MX500, 4TB 7E8, Xonar DGX, XL2730Z + L22e-20
HTPC: i5-2500K, DH67GD, 6GB, GT 1030, 250GB MX500, 1.5TB Barracuda, Xonar DX, G2420HDB
Laptop: MacBook6,1
 
Waco
Gold subscriber
Grand Gerbil Poohbah
Posts: 3250
Joined: Tue Jan 20, 2009 4:14 pm
Location: Los Alamos, NM

Re: Blackblaze's 2018 HDD stats

Wed Mar 27, 2019 5:51 pm

I'm not sure comparing *any* 3.5" drive with a 15K RPM drive is going to be a fair comparison. The standards on those drives were crazy good compared to regular bulk 3.5" drives of any type (including SAS/NL-SAS).
Desktop: X570 Gaming X | 3900X | 32 GB | Alphacool Eisblock Radeon VII | Heatkiller R3 | Samsung 4K 40" | 1 TB SX8200 Pro + 2 TB 660p + 2 TB SATA SSD
NAS: 1950X | Designare EX | 32 GB ECC | 7x8 TB RAIDZ2 | 8x2 TB RAID10 | FreeNAS | ZFS | LSI SAS
 
Krogoth
Gerbil Elder
Posts: 5846
Joined: Tue Apr 15, 2003 3:20 pm
Location: somewhere on Core Prime
Contact:

Re: Blackblaze's 2018 HDD stats

Wed Mar 27, 2019 6:42 pm

Chrispy_ wrote:
The problem is that increased RPM doesn't translate linearly to more performance. Higher RPM drives typically have lower areal density so whilst the lifespan/power/noise increases are linear (or even exponentially worse) with increased RPM, the performance gains are less than linear.

In theory a 7200rpm drive is 33% faster than a 5400rpm drive but in practice the real advantage is often 15% or less.


Increasing RPM has little impact on I/O throughput. It is far more depended on areal density. 10K+ RPM HDDs are forced to use smaller, less dense platters which is why their STR performance has been on par with mainstream 7200RPM units for years. The main advantage of 10K+ HDDs is their access speed which has completely usurped by solid state media.
Gigabyte Z390 AORUS-PRO Coffee Lake R 9700K, 2x8GiB of G.Skill DDR4-3600, Sapphire RX Vega 64, Corsair CX-750M V2 and Fractal Define R4 (W)
Ivy Bridge 3570K, 2x4GiB of G.Skill RIPSAW DDR3-1600, Gigabyte Z77X-UD3H, Corsair CX-750M V2, and PC-7B
 
Chrispy_
Maximum Gerbil
Posts: 4667
Joined: Fri Apr 09, 2004 3:49 pm
Location: Europe, most frequently London.

Re: Blackblaze's 2018 HDD stats

Thu Mar 28, 2019 8:53 am

Krogoth wrote:
Increasing RPM has little impact on I/O throughput. It is far more depended on areal density. 10K+ RPM HDDs are forced to use smaller, less dense platters which is why their STR performance has been on par with mainstream 7200RPM units for years. The main advantage of 10K+ HDDs is their access speed which has completely usurped by solid state media.

Exactly - that's the point I'm trying to make. Spindle speed increases have to be ofset with areal density decreases because of the increased vibration and higher velocities making it harder for the heads to track accurately enough. The benefits of 10K and 15K SAS drives (I still have plenty in use, repurposed away from their original high-IOPS roles) were for reduced seek times and access latency. Short-stroking 15K SAS drives in a striped array was the only way to run an Exchange database for a thousand users without holdups.

These days, anyone professional not using NAND or Optane for high-IOPS, low-latency requirements should be fired on the spot, or at least given the budget they should have been begging for. The main advantage of high-RPM drives has been utterly eclipsed by the 2-3 order-of-magnitude paradigm shift to SSDs. The only mechanical storage I've purchased in four years has been focused on power-consumption and capacity density - so low-RPM, low-power, high-capacity NL-SAS or even plain old SATA drives.
Congratulations, you've noticed that this year's signature is based on outdated internet memes; CLICK HERE NOW to experience this unforgettable phenomenon. This sentence is just filler and as irrelevant as my signature.
 
Glorious
Gold subscriber
Gerbilus Supremus
Posts: 11913
Joined: Tue Aug 27, 2002 6:35 pm

Re: Blackblaze's 2018 HDD stats

Thu Mar 28, 2019 9:26 am

Krogoth wrote:
Increasing RPM has little impact on I/O throughput. It is far more depended on areal density. 10K+ RPM HDDs are forced to use smaller, less dense platters which is why their STR performance has been on par with mainstream 7200RPM units for years. The main advantage of 10K+ HDDs is their access speed which has completely usurped by solid state media.


The smaller size affects the density of the drive (assuming same number of platters+sides), but not areal density. The reason that 10k+ drives had dimensionally smaller platters was related to issues like rigidity and so forth: 15k RPMs is a lot for a 3.5" platter to take.

And it was not "on par" with 7200 RPM drives either: Read any review of the Raptor drives (including the ones here) and while they were obviously significantly faster on Random I/O, they were still substantially faster on sequential I/O.

Some enterprise drives might have longer development pipelines and relied on proven designs instead of the cutting edge when it came to areal density, this is true, but there also wasn't much pressure towards it either because most of their customers use cases were primarily concerned with the much larger access time advantage.

Either way, it wasn't an inherent limitation. The problem was the mechanical physics of platter size, not really any problem with their areal density.
 
Krogoth
Gerbil Elder
Posts: 5846
Joined: Tue Apr 15, 2003 3:20 pm
Location: somewhere on Core Prime
Contact:

Re: Blackblaze's 2018 HDD stats

Thu Mar 28, 2019 9:39 am

Glorious wrote:
Krogoth wrote:
Increasing RPM has little impact on I/O throughput. It is far more depended on areal density. 10K+ RPM HDDs are forced to use smaller, less dense platters which is why their STR performance has been on par with mainstream 7200RPM units for years. The main advantage of 10K+ HDDs is their access speed which has completely usurped by solid state media.


The smaller size affects the density of the drive (assuming same number of platters+sides), but not areal density. The reason that 10k+ drives had dimensionally smaller platters was related to issues like rigidity and so forth: 15k RPMs is a lot for a 3.5" platter to take.

And it was not "on par" with 7200 RPM drives either: Read any review of the Raptor drives (including the ones here) and while they were obviously significantly faster on Random I/O, they were still substantially faster on sequential I/O.

Some enterprise drives might have longer development pipelines and relied on proven designs instead of the cutting edge when it came to areal density, this is true, but there also wasn't much pressure towards it either because most of their customers use cases were primarily concerned with the much larger access time advantage.

Either way, it wasn't an inherent limitation. The problem was the mechanical physics of platter size, not really any problem with their areal density.


Raptors were only faster at STR than mainstream 7200RPM SKUs for a few years until the aforementioned 7200RPM units caught up. That's partly why WD discontinued the Raptor line-up (They were really just excessive enterprise SKUs rebranded). Outside of access speed, 10K RPM HDDs at best are marginally faster than their 7200PRM brethren in the majority of workloads.
Gigabyte Z390 AORUS-PRO Coffee Lake R 9700K, 2x8GiB of G.Skill DDR4-3600, Sapphire RX Vega 64, Corsair CX-750M V2 and Fractal Define R4 (W)
Ivy Bridge 3570K, 2x4GiB of G.Skill RIPSAW DDR3-1600, Gigabyte Z77X-UD3H, Corsair CX-750M V2, and PC-7B
 
Waco
Gold subscriber
Grand Gerbil Poohbah
Posts: 3250
Joined: Tue Jan 20, 2009 4:14 pm
Location: Los Alamos, NM

Re: Blackblaze's 2018 HDD stats

Thu Mar 28, 2019 9:56 am

Krogoth wrote:
Raptors were only faster at STR than mainstream 7200RPM SKUs for a few years until the aforementioned 7200RPM units caught up. That's partly why WD discontinued the Raptor line-up (They were really just excessive enterprise SKUs rebranded). Outside of access speed, 10K RPM HDDs at best are marginally faster than their 7200PRM brethren in the majority of workloads.

They've stagnated in the past 5 years or so because there's zero reason to do R&D on increasing areal density on 10k/15k drives. They're purely in maintenance mode for all of them - if they were pushing the same head/platter tech into the 10k/15k RPM drives the sustained throughput would scale linearly for streaming workloads. Manufacturers are only still making them because they're still selling, somehow.

A 15K RPM drive on modern tech points would be exceeding 500 MB/s streaming...but nobody buys high RPM drives for anything these days except legacy applications. IOPs are for SSDs. Streaming is for SSDs. Bulk storage is slow spinning rust (or "fast" spinning rust with dual actuator drives, those also can do > 500 MB/s streaming).
Desktop: X570 Gaming X | 3900X | 32 GB | Alphacool Eisblock Radeon VII | Heatkiller R3 | Samsung 4K 40" | 1 TB SX8200 Pro + 2 TB 660p + 2 TB SATA SSD
NAS: 1950X | Designare EX | 32 GB ECC | 7x8 TB RAIDZ2 | 8x2 TB RAID10 | FreeNAS | ZFS | LSI SAS
 
Krogoth
Gerbil Elder
Posts: 5846
Joined: Tue Apr 15, 2003 3:20 pm
Location: somewhere on Core Prime
Contact:

Re: Blackblaze's 2018 HDD stats

Thu Mar 28, 2019 10:03 am

Waco wrote:
Krogoth wrote:
Raptors were only faster at STR than mainstream 7200RPM SKUs for a few years until the aforementioned 7200RPM units caught up. That's partly why WD discontinued the Raptor line-up (They were really just excessive enterprise SKUs rebranded). Outside of access speed, 10K RPM HDDs at best are marginally faster than their 7200PRM brethren in the majority of workloads.

They've stagnated in the past 5 years or so because there's zero reason to do R&D on increasing areal density on 10k/15k drives. They're purely in maintenance mode for all of them - if they were pushing the same head/platter tech into the 10k/15k RPM drives the sustained throughput would scale linearly for streaming workloads. Manufacturers are only still making them because they're still selling, somehow.

A 15K RPM drive on modern tech points would be exceeding 500 MB/s streaming...but nobody buys high RPM drives for anything these days except legacy applications. IOPs are for SSDs. Streaming is for SSDs. Bulk storage is slow spinning rust (or "fast" spinning rust with dual actuator drives, those also can do > 500 MB/s streaming).


The physics are more of a problem than the simple lack of R&D budget/interest. It is starting to become difficult for 7200RPM units to reliably read/write data on modern platters. That's kinda why RPMs have been going down on massive HDDs and there's a shift to add more platters. HAMR/MAMR are still terra incognita as far as limitations are concerned.
Gigabyte Z390 AORUS-PRO Coffee Lake R 9700K, 2x8GiB of G.Skill DDR4-3600, Sapphire RX Vega 64, Corsair CX-750M V2 and Fractal Define R4 (W)
Ivy Bridge 3570K, 2x4GiB of G.Skill RIPSAW DDR3-1600, Gigabyte Z77X-UD3H, Corsair CX-750M V2, and PC-7B
 
Waco
Gold subscriber
Grand Gerbil Poohbah
Posts: 3250
Joined: Tue Jan 20, 2009 4:14 pm
Location: Los Alamos, NM

Re: Blackblaze's 2018 HDD stats

Thu Mar 28, 2019 10:10 am

Krogoth wrote:
The physics are more of a problem than the simple lack of R&D budget/interest. It is starting to become difficult for 7200RPM units to reliably read/write data on modern platters. That's kinda why RPMs have been going down on massive HDDs and there's a shift to add more platters. HAMR/MAMR are still terra incognita as far as limitations are concerned.

Let's just say I'm pretty confident that you're wrong on all counts here. :)
Desktop: X570 Gaming X | 3900X | 32 GB | Alphacool Eisblock Radeon VII | Heatkiller R3 | Samsung 4K 40" | 1 TB SX8200 Pro + 2 TB 660p + 2 TB SATA SSD
NAS: 1950X | Designare EX | 32 GB ECC | 7x8 TB RAIDZ2 | 8x2 TB RAID10 | FreeNAS | ZFS | LSI SAS
 
Glorious
Gold subscriber
Gerbilus Supremus
Posts: 11913
Joined: Tue Aug 27, 2002 6:35 pm

Re: Blackblaze's 2018 HDD stats

Thu Mar 28, 2019 10:10 am

2003 the first one is substantially faster at sequential than any 7200rpm competitor

https://www.storagereview.com/western_d ... or_wd360gd

2006 the next one is substantially faster at sequential than any 7200rpm competitor

https://techreport.com/review/10152/wes ... d-drive/12

2009 the next one is substantially faster at sequential than any 7200rpm competitor

https://www.storagereview.com/western_d ... wd3000blfs

2012 It's faster than the fastest 7200rpm, but just barely. At this point, though, they weren't even really trying to push areal density, so that singular drive, that Seagate Barracuda, was significantly more dense (they both had three platters and while the raptor's platters were dimensionally smaller, that's only an advantage of like a half) and, oh, maybe cutting edge isn't so great? Yes, they had market-leading areal density, but uh, the ST3000DM001 might be the only hard drive specific model with its own wikipedia article, and that's not because it PERFORMED well.

https://techreport.com/review/22794/wes ... rd-drive/3

Krogoth wrote:
Raptors were only faster at STR than mainstream 7200RPM SKUs for a few years until the aforementioned 7200RPM units caught up. That's partly why WD discontinued the Raptor line-up (They were really just excessive enterprise SKUs rebranded). Outside of access speed, 10K RPM HDDs at best are marginally faster than their 7200PRM brethren in the majority of workloads.


This simply isn't true, as I've documented. It wasn't a few years, it was almost a decade, and it was only close at the very end in relation to one model. That one model is a legendary reliability disaster, so it's almost not even a true comparison. Every raptor I've ever owned (three! Including the first one!) STILL WORKS. They are all over a decade old. Two of them were actually running earlier this year.

I had a ST3000DM001. I don't think it lasted a full year.

Anyway, WD discontinued the Raptor line because it no place anymore-- SSDs, dude, SSDS. There was no point, and they had clearly stop trying to be competitive on areal density before that. It wasn't worth the engineering investment. The last drive cost like 350, and when you could get a decent 128GB SSD and decent 7200rpm 1 TB drive, what was point?

There really wasn't any. Which is why the last raptor I bought was the 150GB one.



Look, you were wrong. That happens when you made broad announcements like this without even bothering to check first.

Some of those reviews are on this site, and I think you were here when they were written.

Can you just ... stop?
 
Waco
Gold subscriber
Grand Gerbil Poohbah
Posts: 3250
Joined: Tue Jan 20, 2009 4:14 pm
Location: Los Alamos, NM

Re: Blackblaze's 2018 HDD stats

Thu Mar 28, 2019 10:16 am

*not to mention the fact that even "old" 10K/15K RPM drives currently sold still eclipse the fastest 7200 RPM drives you can buy in terms of streaming speed.
Desktop: X570 Gaming X | 3900X | 32 GB | Alphacool Eisblock Radeon VII | Heatkiller R3 | Samsung 4K 40" | 1 TB SX8200 Pro + 2 TB 660p + 2 TB SATA SSD
NAS: 1950X | Designare EX | 32 GB ECC | 7x8 TB RAIDZ2 | 8x2 TB RAID10 | FreeNAS | ZFS | LSI SAS
 
Glorious
Gold subscriber
Gerbilus Supremus
Posts: 11913
Joined: Tue Aug 27, 2002 6:35 pm

Re: Blackblaze's 2018 HDD stats

Thu Mar 28, 2019 10:23 am

Krogoth wrote:
The physics are more of a problem than the simple lack of R&D budget/interest. It is starting to become difficult for 7200RPM units to reliably read/write data on modern platters. That's kinda why RPMs have been going down on massive HDDs and there's a shift to add more platters. HAMR/MAMR are still terra incognita as far as limitations are concerned.


Why should I trust your handwaving like this when you are 100% unreliable on matters anyone can factually check?

Especially when you commented on BOTH of the techreport articles I cited, and you specifically even said this:

Krogoth wrote:
STR is simply how much data is transfer over the cable to memory or whatever. Random seek time is how fast the HDD is able to do a quick access to data a.k.a latency. 7200RPM HDDs need a couple of improvements in areal density if they ever hope of beating the new Raptors in random seek time.


That's not what STR is, and if it is (you define your own terms,whee!),OK, uh... everything you've said here is meaningless, then?

And Areal Density.... uh, dude, the biggest components to random seek time are rotational latency (wait for the disk to spin around to the right spot) and actuator latency (wait for the head to move to the right "track" of the disk), BY FAR. "transfer time" is like an order of magnitude less than either of those.

A "couple of improvements?"

Come on.
 
Waco
Gold subscriber
Grand Gerbil Poohbah
Posts: 3250
Joined: Tue Jan 20, 2009 4:14 pm
Location: Los Alamos, NM

Re: Blackblaze's 2018 HDD stats

Thu Mar 28, 2019 10:26 am

Glorious wrote:
Krogoth wrote:
The physics are more of a problem than the simple lack of R&D budget/interest. It is starting to become difficult for 7200RPM units to reliably read/write data on modern platters. That's kinda why RPMs have been going down on massive HDDs and there's a shift to add more platters. HAMR/MAMR are still terra incognita as far as limitations are concerned.


Why should I trust your handwaving like this when you are 100% unreliable on matters anyone can factually check?

Especially when you commented on BOTH of the techreport articles I cited, and you specifically even said this:

Krogoth wrote:
STR is simply how much data is transfer over the cable to memory or whatever. Random seek time is how fast the HDD is able to do a quick access to data a.k.a latency. 7200RPM HDDs need a couple of improvements in areal density if they ever hope of beating the new Raptors in random seek time.


That's not what STR is, and if it is (you define your own terms,whee!),OK, uh... everything you've said here is meaningless, then?

And Areal Density.... uh, dude, the biggest components to random seek time are rotational latency (wait for the disk to spin around to the right spot) and actuator latency (wait for the head to move to the right "track" of the disk), BY FAR. "transfer time" is like an order of magnitude less than either of those.

A "couple of improvements?"

Come on.

Yeah, I think he's clearly in waaay over his head on this one. Half of what he says is flat out wrong, the other half seems to be random guesses about why things were done certain ways.
Desktop: X570 Gaming X | 3900X | 32 GB | Alphacool Eisblock Radeon VII | Heatkiller R3 | Samsung 4K 40" | 1 TB SX8200 Pro + 2 TB 660p + 2 TB SATA SSD
NAS: 1950X | Designare EX | 32 GB ECC | 7x8 TB RAIDZ2 | 8x2 TB RAID10 | FreeNAS | ZFS | LSI SAS
 
Chrispy_
Maximum Gerbil
Posts: 4667
Joined: Fri Apr 09, 2004 3:49 pm
Location: Europe, most frequently London.

Re: Blackblaze's 2018 HDD stats

Thu Mar 28, 2019 10:32 am

This has become a willy-waving contest that's of no value whatsoever.

Manufacturers never managed to get linear sequential transfer rate improvements with spindle-speed increases. It's just harder to do for all kinds of reasons that we don't really know for sure and have become totally irrelevant since the dawn of mainstream SSDs.

Back in the 10K vs 15K enterprise days, 15K was not 50% faster than 10K or 100% faster than 7200rpm. All else being equal, Areal density has been lower on higher RPM drives. Why? Who cares. SSD's have made high-rpm drives obsolete and development on them has stalled or stopped entirely.

As for consumer 7200rpm drives? All else being equal they are hotter, noisier, hungrier, and not proportionally faster for their troubles than the equivalent 5400rpm or 5900rpm models. Any arguments about performance when talking mechanical drives is moot because their throughput on mixed-IO data is pitiful and the inner tracks of a 7200rpm drive are significantly slower than the outer tracks of a 5400rpm drive. If you actually want sequential improvements from a mechanical bulk storage drive, just buy a larger capacity so that your data is further out on the platters platters or build a striped array (with or without parity). A 3-drive softRAID-5 is still cheaper than a SATA SSD if you are only concerned with sequential transfer rates and can hit 300-500MB/s depending on which drives are used.
Congratulations, you've noticed that this year's signature is based on outdated internet memes; CLICK HERE NOW to experience this unforgettable phenomenon. This sentence is just filler and as irrelevant as my signature.
 
Glorious
Gold subscriber
Gerbilus Supremus
Posts: 11913
Joined: Tue Aug 27, 2002 6:35 pm

Re: Blackblaze's 2018 HDD stats

Thu Mar 28, 2019 10:39 am

Chrispy_ wrote:
This has become a willy-waving contest that's of no value whatsoever.


Maybe in my last comment, because if someone is going to claim the mantle of "I know what I am talking about" and they, well, don't, this shouldn't be a matter of two blowhards arguing "no I *KNOW*". And so I simply demonstrated that my interlocutor was simply not nearly as knowledgeable as he falsely suggests.

But that's not how it started.

I cited numerous reviews, including two from this very site, that contradicted a false technical assertion.

I'm here because knowledgeable people talk about technical things, intelligently.

So if citing the actual results (i.e. the REPORT) of the technical investigations (i.e. the TECH) of this site in this site's own forums is "Willy-waving", quite literally I must ask you:

What is the purpose of any of this?

Because if that's not the purpose of all of this, who is really "willy-waving" here?

EDIT: To be clear, it is sort of ridiculous of you to say everyone is just "willy-waving" and then present an argument about how you, over both of us, have the correct technical line.

I mean, come on. What on earth do you even think you are doing? You cannot be *THAT* self-unaware, can you?
Last edited by Glorious on Thu Mar 28, 2019 10:42 am, edited 1 time in total.
 
Waco
Gold subscriber
Grand Gerbil Poohbah
Posts: 3250
Joined: Tue Jan 20, 2009 4:14 pm
Location: Los Alamos, NM

Re: Blackblaze's 2018 HDD stats

Thu Mar 28, 2019 10:42 am

Glorious wrote:
Because if that's not the purpose of all of this, who is really "willy-waving" here?

I can claim some guilt there. I do get tired of detailing out every single thing, and given it's my day job to know these things, I'm certainly prone to jumping to the "you're wrong" argument.

I apologize.
Desktop: X570 Gaming X | 3900X | 32 GB | Alphacool Eisblock Radeon VII | Heatkiller R3 | Samsung 4K 40" | 1 TB SX8200 Pro + 2 TB 660p + 2 TB SATA SSD
NAS: 1950X | Designare EX | 32 GB ECC | 7x8 TB RAIDZ2 | 8x2 TB RAID10 | FreeNAS | ZFS | LSI SAS
 
Glorious
Gold subscriber
Gerbilus Supremus
Posts: 11913
Joined: Tue Aug 27, 2002 6:35 pm

Re: Blackblaze's 2018 HDD stats

Thu Mar 28, 2019 10:45 am

Waco wrote:
I can claim some guilt there. I do get tired of detailing out every single thing, and given it's my day job to know these things, I'm certainly prone to jumping to the "you're wrong" argument.


I'd rather hear from someone who does this as a day-job than from someone free-form fantasying on the internet. Especially when they've actually commented on the very articles here that completely undermine what they are technically claiming now.

And, as my edit suggests, I fundamentally reject the ridiculous notion that someone who wades in declaring "Your argument has no value, you're both just trying to get over. OH BY THE WAY, HERE IS MY CORRECT SUMMARY OF THE SITUATION THAT DOESN'T EVEN CITE ANYTHING YOU'VE SAID."

Please.
 
Waco
Gold subscriber
Grand Gerbil Poohbah
Posts: 3250
Joined: Tue Jan 20, 2009 4:14 pm
Location: Los Alamos, NM

Re: Blackblaze's 2018 HDD stats

Thu Mar 28, 2019 10:48 am

Agreed on all points. :)
Desktop: X570 Gaming X | 3900X | 32 GB | Alphacool Eisblock Radeon VII | Heatkiller R3 | Samsung 4K 40" | 1 TB SX8200 Pro + 2 TB 660p + 2 TB SATA SSD
NAS: 1950X | Designare EX | 32 GB ECC | 7x8 TB RAIDZ2 | 8x2 TB RAID10 | FreeNAS | ZFS | LSI SAS
 
Glorious
Gold subscriber
Gerbilus Supremus
Posts: 11913
Joined: Tue Aug 27, 2002 6:35 pm

Re: Blackblaze's 2018 HDD stats

Thu Mar 28, 2019 10:51 am

In other words,

If you want to be a referee and get to say that I fouled Krogoth's layup, you don't get to then dunk on us and demand that we sit quietly as you demonstrate how a three pointer should be correctly performed.

Because that's really obnoxious, and I'm not sure why I'm supposed to just accept the argument that I'm showboating and that nothing I did on the court matters while the referee runs back and forth trampoline dunking on both hoops because everyone else has been disqualified and "he'll just the resolve the game for us".
 
Glorious
Gold subscriber
Gerbilus Supremus
Posts: 11913
Joined: Tue Aug 27, 2002 6:35 pm

Re: Blackblaze's 2018 HDD stats

Thu Mar 28, 2019 10:58 am

Chrispy_ wrote:
Manufacturers never managed to get linear sequential transfer rate improvements with spindle-speed increases. It's just harder to do for all kinds of reasons that we don't really know for sure and have become totally irrelevant since the dawn of mainstream SSDs.


I mean, I'm totally baffled, because I just posted a whole series of high-quality reviews that demonstrate, EMPIRICALLY, that Western Digital did EXACTLY THAT between the years of 2003 and 2012.

You, however, ignore all that while chastising me and say that they "never" managed to do that.

:blink:

Is your retort to that well-sourced claim literally just that I'm waving my dick around, so shut up, here's Chrispy_ with a completely uncited list of "true facts"?

Actually, YES! It was! :o I mean, not a retort because you clearly didn't read literally any of it before declaring it had no value, so I'm actually being unjustifiably kind in characterizing it as such, but yeah, that's what just happened.
 
Usacomp2k3
Gerbil God
Posts: 23043
Joined: Thu Apr 01, 2004 4:53 pm
Location: Orlando, FL
Contact:

Re: Blackblaze's 2018 HDD stats

Thu Mar 28, 2019 11:41 am

Would there be value in a lower spindle speed drive? Would that allow for more density? For my NAS Drives, anything faster than gigabit is a waste. I'd much rather trade that for higher density, lower cost, or higher reliability. Most 2.5" were ~3900-7200rpm while the 3.5" were ~5400-10,000. Would a 4,000rpm 3.5" drive that give me any of the above improvements?
 
Chrispy_
Maximum Gerbil
Posts: 4667
Joined: Fri Apr 09, 2004 3:49 pm
Location: Europe, most frequently London.

Re: Blackblaze's 2018 HDD stats

Thu Mar 28, 2019 11:42 am

Glorious wrote:
I mean, I'm totally baffled, because I just posted a whole series of high-quality reviews that demonstrate, EMPIRICALLY, that Western Digital did EXACTLY THAT between the years of 2003 and 2012.
You, however, ignore all that while chastising me and say that they "never" managed to do that.
:blink:
Is your retort to that well-sourced claim literally just that I'm waving my dick around, so shut up, here's Chrispy_ with a completely uncited list of "true facts"?
Actually, YES! It was! :o I mean, not a retort because you clearly didn't read literally any of it before declaring it had no value, so I'm actually being unjustifiably kind in characterizing it as such, but yeah, that's what just happened.


You're 100% right, I didn't read your links because I didn't have to. My memory worked absolutely fine - but since you threw a hissyfit I'll pull out the relevant information from your links:

https://www.storagereview.com/western_d ... or_wd360gd
10K Raptor 63.1MB/s, vs 7200rpm Maxtor 59.2MB/s - so 39% more spindle speed gives only 7% more throughput.

https://techreport.com/review/10152/wes ... d-drive/12
10K Raptor 77.5MB/s, vs Barracuda 7200 66.6MB/s - so 39% more spindle speed gives only 16% more throughput.

https://www.storagereview.com/western_d ... wd3000blfs
10K Raptor 119MB/s, vs Barracude ES.2 104MB/s - so 39% more spindle speed gives only 14% more throughput

https://techreport.com/review/22794/wes ... rd-drive/3
10K Raptor 165MB/s, vs Barracuda 3TB 163MB/s - 39% more spindle speed gives only 1% more throughput.

What are you trying to prove? I'm saying that higher spindle speed does not equate to performance linearly, as proven by your links across several years.

If you're trying to compare a 1TB WD Blue with a 1TB WD Raptor, you've just digging yourself a deeper hole since one is a single-platter cut-down 3.5" drive for product-segmentation and the other is a multi-platter 2.5" drive. They are not remotely comparable - at the very least you have to match multi-platter against multi-platter if you're talking about sequential performance, areal density, and spindle speed.
Last edited by Chrispy_ on Thu Mar 28, 2019 11:49 am, edited 1 time in total.
Congratulations, you've noticed that this year's signature is based on outdated internet memes; CLICK HERE NOW to experience this unforgettable phenomenon. This sentence is just filler and as irrelevant as my signature.
 
Chrispy_
Maximum Gerbil
Posts: 4667
Joined: Fri Apr 09, 2004 3:49 pm
Location: Europe, most frequently London.

Re: Blackblaze's 2018 HDD stats

Thu Mar 28, 2019 11:46 am

Usacomp2k3 wrote:
Would there be value in a lower spindle speed drive? Would that allow for more density? For my NAS Drives, anything faster than gigabit is a waste. I'd much rather trade that for higher density, lower cost, or higher reliability. Most 2.5" were ~3900-7200rpm while the 3.5" were ~5400-10,000. Would a 4,000rpm 3.5" drive that give me any of the above improvements?


Yes, but it's almost irrelevant since the drives that allow variable, configurable spindle speeds lower than 5400rpm are enterprise drives like the ones I have in my NASes and they each cost silly money for consumer NAS drives, as well as being overkill for low-RPM operation since they're scalable to 7200rpm anyway.

The quietest, densest drives are going to be helium-filled 5400rpm models. I believe the WD/HGST-rebranded WD Red 8TB is in this category. Reviews comparing multiple mechanical drives on the same testbed these days are few and far between though :(
Congratulations, you've noticed that this year's signature is based on outdated internet memes; CLICK HERE NOW to experience this unforgettable phenomenon. This sentence is just filler and as irrelevant as my signature.
 
Glorious
Gold subscriber
Gerbilus Supremus
Posts: 11913
Joined: Tue Aug 27, 2002 6:35 pm

Re: Blackblaze's 2018 HDD stats

Thu Mar 28, 2019 11:56 am

chrispy_ wrote:
You're right, I didn't read your links because I didn't have to, but since you threw a hissyfit I'll pull out the relevant information from them:


"I didn't read your links because I didn't have to"

But I'm the one dick-waving?

---

And, wow! Look, throwing a hissy-fit gets you to actually engage with what I've said, that's progress from before!

Chrispy_ wrote:
What are you trying to prove? I'm saying that higher spindle speed does not equate to performance linearly, as proven by your links across several years.


I'll admit that I misread you, because you were making a statement unrelated to my dispute with Krogoth.

You didn't mean linear as an synonym for sequential, and hence what you said wasn't redundant. I understand now, pursuant to this and your previous remarks, that you meant "linear" in relation to how performance didn't scale.

I get that now, but since it was never something I disputed, I didn't take your meaning. I made the mistake of thinking that you were perhaps commenting on what was actually being discussed instead of something I never refuted or disagreed with.

EDIT: I would like to think that'd you actually understand this was a legitimate reading of your sentence (if not your whole post, which did later make it clear contextually---I also admit that I didn't read any further, because I was miffed you didn't really read me). You didn't say scale, and my parsing can be correct if we accept a little redundant weirdness with linear sequential.

But maybe you won't. Whatever.

Chrispy_ wrote:
If you're trying to compare a 1TB WD Blue with a 1TB WD Raptor, you've just digging yourself a deeper hole since one is a single-platter cut-down 3.5" drive for product-segmentation and the other is a multi-platter 2.5" drive. At the very least you have to match multi-platter against multi-platter.


....and so you've returned the favor by completely misreading me, because I've already blatantly demonstrated I understand the importance of matching multi-platter for multi-platter and I actually have no what you think I'm "trying to do"

---

So THANK YOU for adding SO MUCH VALUE to the conversation as well as not "WILLY-WAVING" yourself.

You totally helped this dispute and just cleared everything up.
 
Waco
Gold subscriber
Grand Gerbil Poohbah
Posts: 3250
Joined: Tue Jan 20, 2009 4:14 pm
Location: Los Alamos, NM

Re: Blackblaze's 2018 HDD stats

Thu Mar 28, 2019 12:00 pm

Chrispy_ wrote:
What are you trying to prove? I'm saying that higher spindle speed does not equate to performance linearly, as proven by your links across several years.

At the same tech level, spindle speed scales streaming speed linearly. That doesn't necessarily translate to consumer drives like the Raptor, even if they shared DNA with the enterprise drives. You're also comparing 7200 RPM 3.5" models with 10K RPM 2.5" models. The linear speed of the heads over the platters at a given RPM very much favors the 3.5" drives.

If you look at the transfer rates at the insides of the 3.5" platters (where you can actually compare directly to 2.5" drives) the performance scales almost exactly linearly with rotation speed.
Desktop: X570 Gaming X | 3900X | 32 GB | Alphacool Eisblock Radeon VII | Heatkiller R3 | Samsung 4K 40" | 1 TB SX8200 Pro + 2 TB 660p + 2 TB SATA SSD
NAS: 1950X | Designare EX | 32 GB ECC | 7x8 TB RAIDZ2 | 8x2 TB RAID10 | FreeNAS | ZFS | LSI SAS
 
Glorious
Gold subscriber
Gerbilus Supremus
Posts: 11913
Joined: Tue Aug 27, 2002 6:35 pm

Re: Blackblaze's 2018 HDD stats

Thu Mar 28, 2019 12:15 pm

Chrispy_ how many disputes are you going to wade into with this ridiculous inflammatory approach of "POX ON BOTH YOUR HOUSES --- here is what you two should really be arguing about, and my correct take on it. Fin."

Are you out of your mind? How does that help anything?

You are acting like I have no idea what I'm talking about only because I was legitimately confused with your "correction" that had nothing to do with the actual dispute.

---

Let's ignore that, then.

What did your intercession have to with my argument with Krogoth?

Nothing. It has nothing to do with it, except to aggressively assert that YOUR unrelated point completely superseded our actual contention, to extent that our discussion was "valueless". (I mean, it wasn't you, talking to yourself, so yes, ofc :roll: )

It was like you were angry that we were talking to each other about something other than your much earlier post where you opined upon this whole linear scaling bit. How dare the plebs not pay attention to the patricians!

And we're the vain ones?

AGAIN HOW ARE YOU HELPING ANYTHING?

EDIT: I mean look!

Chrispy_ wrote:
The problem is that increased RPM doesn't translate linearly to more performance.



Chrispy_, I heard you the first time. I just didn't have anything to say about it.

Thanks?


I mean, "The problem"--- uh, OK? I guess that's "your" problem?

My problem was that someone here, once again, was saying something that my own personal experience (not to mention the reviews on this site) totally belied.

But, sure, sure, that's "has no value" because *YOU* don't value it.

----

I think I have might a new problem, an actually even more annoying problem, now.

Thanks, Again?
 
Waco
Gold subscriber
Grand Gerbil Poohbah
Posts: 3250
Joined: Tue Jan 20, 2009 4:14 pm
Location: Los Alamos, NM

Re: Blackblaze's 2018 HDD stats

Thu Mar 28, 2019 12:38 pm

Short version:
10K/15K RPM drives are faster, but not fast enough to be useful today except legacy apps. Everyone agrees. Everyone is happy.

The end.
Desktop: X570 Gaming X | 3900X | 32 GB | Alphacool Eisblock Radeon VII | Heatkiller R3 | Samsung 4K 40" | 1 TB SX8200 Pro + 2 TB 660p + 2 TB SATA SSD
NAS: 1950X | Designare EX | 32 GB ECC | 7x8 TB RAIDZ2 | 8x2 TB RAID10 | FreeNAS | ZFS | LSI SAS
  • 1
  • 2
  • 3
  • 4
  • 5
  • 7

Who is online

Users browsing this forum: No registered users and 2 guests
GZIP: On