Jump to content

WD Red speed difference, should I be worried?

I'm currently going through a burn-in for my new drives to make sure they're good before I put them in the NAS and start storing my data on them.

As burn-in procedure, I'm doing a triple pass overwrite in CCleaner followed by a full error check in HDTune. That's enough to work them hard for 30-40 hours, so if one of them has a problem, this should be enough to highlight it.

I started the overwrite on both disks at the same time (or at least with just a few seconds in between). We're 24 hours later and the difference is quite large.

post-139790-0-69183800-1433761856_thumb.

Should I be worried about this difference? No idea what could cause it except for the HDDs themselves. Both are connected via a SATA port on the same on-board controller which is far from saturated. Both instances of CCleaner aren't fighting for resources either, I have 13GB RAM unused and my CPU usage has been steady at 15-ish percent all day.

For the record, these are only the first 2 drives, once they're done I have 6 more of these to go. It's going to be a fun week ...

Link to comment
Share on other sites

Link to post
Share on other sites

 

Spoiler
Spoiler

AMD 5000 Series Ryzen 7 5800X| MSI MAG X570 Tomahawk WiFi | G.SKILL Trident Z RGB 32GB (2 * 16GB) DDR4 3200MHz CL16-18-18-38 | Asus GeForce GTX 3080Ti STRIX | SAMSUNG 980 PRO 500GB PCIe NVMe Gen4 SSD M.2 + Samsung 970 EVO Plus 1TB PCIe NVMe M.2 (2280) Gen3 | Cooler Master V850 Gold V2 Modular | Corsair iCUE H115i RGB Pro XT | Cooler Master Box MB511 | ASUS TUF Gaming VG259Q Gaming Monitor 144Hz, 1ms, IPS, G-Sync | Logitech G 304 Lightspeed | Logitech G213 Gaming Keyboard |

PCPartPicker 

Link to comment
Share on other sites

Link to post
Share on other sites

FTFY, mentioning him is a bit of an issue apparently.

Good idea indeed, thanks !

yeah tagging member with symbols is really hard , just used

 @[member='name']

 

Spoiler
Spoiler

AMD 5000 Series Ryzen 7 5800X| MSI MAG X570 Tomahawk WiFi | G.SKILL Trident Z RGB 32GB (2 * 16GB) DDR4 3200MHz CL16-18-18-38 | Asus GeForce GTX 3080Ti STRIX | SAMSUNG 980 PRO 500GB PCIe NVMe Gen4 SSD M.2 + Samsung 970 EVO Plus 1TB PCIe NVMe M.2 (2280) Gen3 | Cooler Master V850 Gold V2 Modular | Corsair iCUE H115i RGB Pro XT | Cooler Master Box MB511 | ASUS TUF Gaming VG259Q Gaming Monitor 144Hz, 1ms, IPS, G-Sync | Logitech G 304 Lightspeed | Logitech G213 Gaming Keyboard |

PCPartPicker 

Link to comment
Share on other sites

Link to post
Share on other sites

first question, are both HDD's of same age, and how much data writes difference is there between two, if one of the drive is heavily abused before or the other drive is older than the newer, then ofcourse there's gonna be a difference.

 

Spoiler
Spoiler

AMD 5000 Series Ryzen 7 5800X| MSI MAG X570 Tomahawk WiFi | G.SKILL Trident Z RGB 32GB (2 * 16GB) DDR4 3200MHz CL16-18-18-38 | Asus GeForce GTX 3080Ti STRIX | SAMSUNG 980 PRO 500GB PCIe NVMe Gen4 SSD M.2 + Samsung 970 EVO Plus 1TB PCIe NVMe M.2 (2280) Gen3 | Cooler Master V850 Gold V2 Modular | Corsair iCUE H115i RGB Pro XT | Cooler Master Box MB511 | ASUS TUF Gaming VG259Q Gaming Monitor 144Hz, 1ms, IPS, G-Sync | Logitech G 304 Lightspeed | Logitech G213 Gaming Keyboard |

PCPartPicker 

Link to comment
Share on other sites

Link to post
Share on other sites

@Rohith_Kumar_Sp They are all brand new, unopened plastic bag and all.

I didn't check the other 6 yet, I can only attach 2 to the PC at the same time due to lack of SATA ports.

at the end of the day, its your cpu prioritizing CCleaner, try using samsung magician software to check the performance of both for more closer comparison

 

Spoiler
Spoiler

AMD 5000 Series Ryzen 7 5800X| MSI MAG X570 Tomahawk WiFi | G.SKILL Trident Z RGB 32GB (2 * 16GB) DDR4 3200MHz CL16-18-18-38 | Asus GeForce GTX 3080Ti STRIX | SAMSUNG 980 PRO 500GB PCIe NVMe Gen4 SSD M.2 + Samsung 970 EVO Plus 1TB PCIe NVMe M.2 (2280) Gen3 | Cooler Master V850 Gold V2 Modular | Corsair iCUE H115i RGB Pro XT | Cooler Master Box MB511 | ASUS TUF Gaming VG259Q Gaming Monitor 144Hz, 1ms, IPS, G-Sync | Logitech G 304 Lightspeed | Logitech G213 Gaming Keyboard |

PCPartPicker 

Link to comment
Share on other sites

Link to post
Share on other sites

~snip~

 

Hey again fellow Captain :)
 
This might be caused by something in the SATA controller prioritizing one drive or another. you can run a benchmarking tool (multiple different tools would be better) and WD Data Lifeguard Diagnostic tool to check if the drives are healthy and how they perform. They should all perform equally and I doubt that the difference is caused by them (it shouldn't). 320MB for 24 hours is around 3.7 Bytes per second difference (Which I doubt it would show in benchmarks, unless you are looking at the 3rd digit after the dot), if my mind is not deceiving me.
 
Have you encountered a similar thing before with identical drives? 
 
Thanks @Rohith_Kumar_Sp for tagging me, but I still don't get notifications. There's a problem with the user name or something. I've talked with the mods and so far there's nothing they can do. :) Feel free to send me a PM if you need me.
 
Captain_WD.

If this helped you, like and choose it as best answer - you might help someone else with the same issue. ^_^
WDC Representative, http://www.wdc.com/ 

Link to comment
Share on other sites

Link to post
Share on other sites

@Captain_WD This isn't a mere 320MB, the T:\ drive is now about to finish (a minute to go) and the difference is 600GB.

The S:\ drive is only at 85% at this point, so it looks like it's 15% slower.

Never encountered anything like this, the 4 Greens that are in my current EX4 (I know, I know, not supposed to use those) performed identical a few years ago when I did the burn-in on them. That's why I'm so worried.

I'll run CrystalDiskMark and HDtune on both (once the slow one is done with the wipe) and compare them. Fingers crossed ...

Link to comment
Share on other sites

Link to post
Share on other sites

@Captain_WD This isn't a mere 320MB, the T:\ drive is now about to finish (a minute to go) and the difference is 600GB.

The S:\ drive is only at 85% at this point, so it looks like it's 15% slower.

Never encountered anything like this, the 4 Greens that are in my current EX4 (I know, I know, not supposed to use those) performed identical a few years ago when I did the burn-in on them. That's why I'm so worried.

I'll run CrystalDiskMark and HDtune on both (once the slow one is done with the wipe) and compare them. Fingers crossed ...

 I just noticed that on the screenshot, the values are in MB, not B. Apologies. 

With 600GB difference in 24 hours, it's roughly 7MB/s difference between the two drives, which is not normal. Do run the benchmark several times so you have accurate-enough data and see if it's real, or it's caused by something else. 

 

Captain_WD. 

If this helped you, like and choose it as best answer - you might help someone else with the same issue. ^_^
WDC Representative, http://www.wdc.com/ 

Link to comment
Share on other sites

Link to post
Share on other sites

What is your NAS setup going to be out of curiosity? 

System/Server Administrator - Networking - Storage - Virtualization - Scripting - Applications

Link to comment
Share on other sites

Link to post
Share on other sites

@Captain_WD Thanks, will do. The one that's finished is looking good, but that's not the one I'm worried about.

Speedfan has some trouble interpreting the S.M.A.R.T readings, but I think that's more Speedfan's problem than one with the drive. The drive is returning values of exactly 100 on raw read error rate, write error rate and spin up time. Speedfan thinks it should be 255

@Eniqmatic

Case : Fractal Designs Define R5

Case fans : 3x BeQuiet Silent Wings 2 140mm

MoBo : Intel Server S1200V3RPS

CPU : Intel Core I3-4170 (ECC-compatible)

CPU cooler : BeQuiet Shadow Rock 2

RAM : 16GB (2x 8GB) Kingston Valueram ECC 8GB DDR3-1333

RAID card : LSI 9211-8i flashed to IT mode

PSU : BeQuiet Dark Power Pro P9 650W (had this one laying around still)

SSDs : 2x Crucial MX200 250GB (RAID 1, used for regular data, where access/seek time is most important to me)

HDDs : 8x WD Red 4TB (either 16TB RAID 1 or 24TB RAID 6, haven't decided yet but leaning towards the latter right now. Used for video, software installers and ISO files)

OS : FreeNAS, ZFS file system

Link to comment
Share on other sites

Link to post
Share on other sites

@Captain_WD Thanks, will do. The one that's finished is looking good, but that's not the one I'm worried about.

Speedfan has some trouble interpreting the S.M.A.R.T readings, but I think that's more Speedfan's problem than one with the drive. The drive is returning values of exactly 100 on raw read error rate, write error rate and spin up time. Speedfan thinks it should be 255

@Eniqmatic

Case : Fractal Designs Define R5

Case fans : 3x BeQuiet Silent Wings 2 140mm

MoBo : Intel Server S1200V3RPS

CPU : Intel Core I3-4170 (ECC-compatible)

CPU cooler : BeQuiet Shadow Rock 2

RAM : 16GB (2x 8GB) Kingston Valueram ECC 8GB DDR3-1333

RAID card : LSI 9211-8i flashed to IT mode

PSU : BeQuiet Dark Power Pro P9 650W (had this one laying around still)

SSDs : 2x Crucial MX200 250GB (RAID 1, used for regular data, where access/seek time is most important to me)

HDDs : 8x WD Red 4TB (either 16TB RAID 1 or 24TB RAID 6, haven't decided yet but leaning towards the latter right now. Used for video, software installers and ISO files)

OS : FreeNAS, ZFS file system

 

Good little system! Any reason you didn't run the tests from FreeNAS or did you just find it easier to do it from Windows?

 

I assume you mean you'll be using RAIDZ2 rather than RAID 6?

System/Server Administrator - Networking - Storage - Virtualization - Scripting - Applications

Link to comment
Share on other sites

Link to post
Share on other sites

Good little system! Any reason you didn't run the tests from FreeNAS or did you just find it easier to do it from Windows?

 

I assume you mean you'll be using RAIDZ2 rather than RAID 6?

The reason I'm doing it in Windows is because (due to the massive choice in PC parts here) I had to source the motherboard, RAID card and the RAM from Ebay. (Italy, the US and Hong Kong respectively)

It's going to take a week before I have all the parts, and maybe longer because I expect the RAID card to take its sweet time to get through Customs.

Seeing as I have to wait that long, I want to have the HDDs stress-tested already.

And yes, I mean RAIDZ2. RAID 6 is just an easier way to describe it, in case the other person isn't familiar with ZFS (not that I'm that familiar with it, I'm only just starting to get into the whole self-built NAS thing myself).

I didn't notice your signature before, otherwise I would have used the proper term.

Link to comment
Share on other sites

Link to post
Share on other sites

The reason I'm doing it in Windows is because (due to the massive choice in PC parts here) I had to source the motherboard, RAID card and the RAM from Ebay. (Italy, the US and Hong Kong respectively)

It's going to take a week before I have all the parts, and maybe longer because I expect the RAID card to take its sweet time to get through Customs.

Seeing as I have to wait that long, I want to have the HDDs stress-tested already.

And yes, I mean RAIDZ2. RAID 6 is just an easier way to describe it, in case the other person isn't familiar with ZFS (not that I'm that familiar with it, I'm only just starting to get into the whole self-built NAS thing myself).

 

OK cool. You could still effectively run FreeNAS from any system in order to test the drives since it runs from the pen drive. Anyway, this doesn't matter to much.

 

Have you benchmarked the drives on their own yet? A simple test to see how they perform individually, taking everything else out of the equation such as CPU allocation. Did you check your SMART results yet or not?

System/Server Administrator - Networking - Storage - Virtualization - Scripting - Applications

Link to comment
Share on other sites

Link to post
Share on other sites

Only checked the first one, number two is still wiping.

Perhaps I should indeed do a benchmark on each one first and see if that looks good already.

Not sure about running FreeNAS on the main PC, I work mostly at home and need it for work. I really can't take it offline for a day or longer.

Currently the NAS drives are in the R5, but fed by the main PC.

post-139790-0-27025500-1433770857.jpg
Link to comment
Share on other sites

Link to post
Share on other sites

Only checked the first one, number two is still wiping.

Perhaps I should indeed do a benchmark on each one first and see if that looks good already.

 

They should all show close to identical results. If not, contact the place where you got them from and possibly go for an RMA procedure. :)

RAIDZ2 might be the better option (just my humble opinion). :)

 

Captain_WD.

If this helped you, like and choose it as best answer - you might help someone else with the same issue. ^_^
WDC Representative, http://www.wdc.com/ 

Link to comment
Share on other sites

Link to post
Share on other sites

I would, if you get similar results when doing them individually, I would suggest that something else is factoring into the equation. If you don't then RMA it. Or compare it to another drive so you can get a more defined conclusion.

System/Server Administrator - Networking - Storage - Virtualization - Scripting - Applications

Link to comment
Share on other sites

Link to post
Share on other sites

Good your actually doing this, too often I see people chucking drives in without weeding out the bad ones first. If you do have a bad one there then it's a prime example.

 

I did the same with mine, ended up being like a 40 hour burn in test on mine.

System/Server Administrator - Networking - Storage - Virtualization - Scripting - Applications

Link to comment
Share on other sites

Link to post
Share on other sites

Good your actually doing this, too often I see people chucking drives in without weeding out the bad ones first. If you do have a bad one there then it's a prime example.

 

I did the same with mine, ended up being like a 40 hour burn in test on mine.

I recall seeing a graph in one of the Tek's or Wendell's videos (Wendell was talking about how they pulled all their Seagates because they died way too much) that showed that there's about 5% chance of a WD being bad early on. With 8 disks, that means that statistically I have 40% chance that one disk is bad. That's enough to check them all first.

I also had an issue once where I was having one bad Maxtor after another. I'd RMA one every few days for 3 weeks. Turns out they all came from the same truck, which had been in an accident. Stuff like that makes you worry about how easy it is to damage such fragile things in transport, even though I'm sure that WD does take precautions to have the drives protect themselves from transport damage.

Link to comment
Share on other sites

Link to post
Share on other sites

I recall seeing a graph in one of Wendell's videos that showed that there's about 5% chance of a WD being bad early on. With 8 disks, that means that statistically I have 40% chance that one disk is bad. That's enough to check them all first.

I also had an issue once where I was having one bad Maxtor after another. I'd RMA one every few days for 3 weeks. Turns out they all came from the same truck, which had been in an accident. Stuff like that makes you worry about how easy it is to damage such fragile things in transport, even though I'm sure that WD does take precautions to have the drives protect themselves from transport damage.

I haven't seen that, would be interesting to see if you can find it.

 

I had 4 drives in recently (WD RED's) which I did the 40 hours on, no bad disks. Before that I had 5 WD Blacks which I also did the same on and had none either.

 

I think that was common practice a couple years ago was to buy half of your disks from a different retailer or at a different time to make sure you didn't get from the same batch. Don't think people worry about it so much anymore.

 

I would get all the drives out and start doing some benches on them, at least that way it might reveal something rather quickly rather than waiting for as much time as you had for the drive to wipe. Tests can be done within a couple of minutes. If they all come back similar then you could then move onto more demanding tests.

System/Server Administrator - Networking - Storage - Virtualization - Scripting - Applications

Link to comment
Share on other sites

Link to post
Share on other sites

I haven't seen that, would be interesting to see if you can find it.

Just did a quick search through the recent videos and I can't find it anymore. That's the downside of just having random videos on one screen while working on the other.

If I do come across it again, I'll let you know.

EDIT : Found it ! The Tek 177, skip to around 46:30

Link to comment
Share on other sites

Link to post
Share on other sites

Just thought I'd double-post to give an update.

I've benchmarked the 7 drives I have (picking up #8 later today, they only had 7 in stock when I bought them).

The one that I was so worried about performs perfectly normal during reads and sequential writes, but is anywhere from 9 to 23 percent slower than all the others during 4K and 4K QD32 writes.

The disk made it through 30h of consistent 4k writing just fine though. Without airflow, the temps were in the 40-42°C range, which seemed to be the normal temperature for all disks at the end of the tests.

I'm afraid you'll have to read this spreadsheet from the bottom up, I started from the bottom of the drive trays and the spreadsheet's layout is pretty much in the same order as the hard drives.

I ran CrystalDiskMark with only one of the drives connected and on the default settings (5 passes, 1000MB each).

That test was done 5 times, then I let the spreadsheet calculate the average in the grey boxes (so that's an average of 25 tests really) and the percentage compared to the reference drive. Whenever the difference was 10% or more, I highlighted the result in green

I used the bottom HDD's average results as a reference point. The bottom disk incidentally was the one I was so worried about.

post-139790-0-37180700-1433825237_thumb.

The tests show that speeds can be rather inconsistent. I have one drive with a significantly lower sequential speed, and there's on average a 5-10% difference in 512K and 4K speeds, with one drive doing particularly bad being on average around 15% slower than the others.

Conclusion : I guess there's not only a silicon lottery but a platter lottery too. The first one I tested just happened to be the slowest of the bunch. I just hope it doesn't bog the entire system down too much if I need to write small files to it.

Link to comment
Share on other sites

Link to post
Share on other sites

Snip

 

Good work.

 

You won't notice in RAIDZ2, you will be bogged down by your network for sure.

 

How was the SMART, all good I take it?

System/Server Administrator - Networking - Storage - Virtualization - Scripting - Applications

Link to comment
Share on other sites

Link to post
Share on other sites

SMART was perfect for all drives, but of course only 2 of them have gone through the burn-in procedure so far. The next 2 have about 18 hours of wiping to go.

I'm curious how it'll perform, I've looked for RAIDZ2 benchmarks but they are all over the place. If my network is the bottleneck I'll be more than happy.

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×