Hitachi Deskstar 5k3000 vs Samsung F4s

cnick79

[H]ard|Gawd
Joined
Oct 5, 2004
Messages
1,836
How do these Hitachi 2TB Deskstars compare to the 2TB Samsung F4? The Hitachi Drvies can be had for $79.99 with promo code (EMCKHKJ26) and those Sammy's are always dropping to the same price.

I'm looking to purchase 4 at the moment so which ones would be good for a ZFS system?
 
Here is a comparison someone did on 2ch between the Hitachi 2TB 5K3000, the 2TB Samsung F4, and the 2TB Western Digital Green.

The Hitachi 5K3000 appears to have similar sequential performance compared to the Samsung F4, but the 5K3000 has the clear advantage in 4KB random reads/writes.

2ch.net said:
●Hitachi HDS5C3020ALA632 5K3000 (667GB Platter*3)
5944 rpm
【HD Tune】
Transfer Rate Minimum : 61.8 MB/sec
Transfer Rate Maximum : 143.7 MB/sec
Transfer Rate Average : 102.3 MB/sec
Access Time : 15.8 ms
Burst Rate : 169.4 MB/sec
CPU Usage : -1.0%

【Crystal Disk Mark 3.0】 (Test Size : 1000 MB)
Sequential Read : 137.446 MB/s
Sequential Write : 134.952 MB/s
Random Read 512KB : 49.940 MB/s
Random Write 512KB : 52.001 MB/s
Random Read 4KB (QD=1) : 0.632 MB/s [ 154.2 IOPS]
Random Write 4KB (QD=1) : 1.660 MB/s [ 405.3 IOPS]
Random Read 4KB (QD=32) : 1.625 MB/s [ 396.7 IOPS]
Random Write 4KB (QD=32) : 0.908 MB/s [ 221.6 IOPS]

●SAMSUNG HD204UI (667GB Platter*3)
5266 rpm
【HD Tune】
Transfer Rate Minimum : 50.6 MB/sec
Transfer Rate Maximum : 143.5 MB/sec
Transfer Rate Average : 103.1 MB/sec
Access Time : 16.0 ms
Burst Rate : 196.6 MB/sec
CPU Usage : 3.0%

【Crystal Disk Mark 3.0】 (Test Size : 1000 MB)
Sequential Read : 138.3 MB/s
Sequential Write : 135.2 MB/s
Random Read 512KB : 39.83 MB/s
Random Write 512KB : 55.65 MB/s
Random Read 4KB : 0.454 MB/s
Random Write 4KB : 1.005 MB/s
Random Read 4KB QD32 : 0.557 MB/s
Random Write 4KB QD32 : 1.007 MB/s

●WD20EARS-00MVWB0 (667GB Platter*3)
5001 rpm
【HD Tune】
Transfer Rate Minimum : 51.4 MB/sec
Transfer Rate Maximum : 122.1 MB/sec
Transfer Rate Average : 93.1 MB/sec
Access Time : 13.4 ms
Burst Rate : 165.7 MB/sec
CPU Usage : 1.0%

【Crystal Disk Mark 3.0】 (Test Size : 1000 MB)
Sequential Read : 121.0 MB/s
Sequential Write : 118.9 MB/s
Random Read 512KB : 41.66 MB/s
Random Write 512KB : 64.05 MB/s
Random Read 4KB : 0.532 MB/s
Random Write 4KB : 0.884 MB/s
Random Read 4KB QD32 : 0.556 MB/s
Random Write 4KB QD32 : 1.032 MB/s

====== HddRpmEst v0.1.5 === Report ============================
Target HDD       : Hitachi HDS5C3020ALA632 5K3000(667GB Platter*3)

Rotational Speed (estimated)     : 5944 rpm
Average Access Time  : 25.2 msec
Average Access Time (10% Short Stroke)  : 13.8 msec
Transfer Rate(Outer,Maximum)  : sustained 152.8 MB/s / burst 165.0 MB/s = 92.6 %
Transfer Rate(Outer,Average)  : sustained 137.4 MB/s / burst 149.4 MB/s = 92.0 %
Transfer Rate(Inner,Average)  : sustained 65.2 MB/s / burst 70.4 MB/s = 92.6 %
Transfer Rate(Inner,Minimum)  : sustained 62.1 MB/s / burst 67.0 MB/s = 92.7 %
Inner/Outer Ratio(Average,Smallest) :  47.5 %, 40.6 %  /    47.1 %, 40.6 %

====== HddRpmEst v0.1.5 === Report ===========================
Target HDD       : SAMSUNG HD204UI(667GB Platter*3)

Rotational Speed (estimated) : 5266 rpm
Average Access Time  : 23.2 msec
Average Access Time (10% Short Stroke)  : 15.1 msec
Transfer Rate(Outer,Maximum)  : sustained 144.3 MB/s / burst 163.4 MB/s = 88.3 %
Transfer Rate(Outer,Average)  : sustained 142.0 MB/s / burst 162.6 MB/s = 87.3 %
Transfer Rate(Inner,Average)  : sustained 60.3 MB/s / burst 68.0 MB/s = 88.7 %
Transfer Rate(Inner,Minimum)  : sustained 60.3 MB/s / burst 67.8 MB/s = 88.9 %
Inner/Outer Ratio(Average,Smallest) :  42.5 %, 41.8 %  /    41.8 %, 41.5 %

====== HddRpmEst v0.1.5 === Report =========================
Target HDD       : WDC WD20EARS-00MVWB0 (667GB Platter*3)

Rotational Speed (estimated)     : 5001 rpm
Average Access Time  : 16.0 msec
Average Access Time (10% Short Stroke)  : 11.5 msec
Transfer Rate(Outer,Maximum)  : sustained 127.8 MB/s / burst 145.6 MB/s = 87.8 %
Transfer Rate(Outer,Average)  : sustained 124.4 MB/s / burst 143.0 MB/s = 87.0 %
Transfer Rate(Inner,Average)  : sustained 56.8 MB/s / burst 64.7 MB/s = 87.8 %
Transfer Rate(Inner,Minimum)  : sustained 53.3 MB/s / burst 60.8 MB/s = 87.7 %
Inner/Outer Ratio(Average,Smallest) :  45.7 %, 41.7 %  /    45.2 %, 41.8 %


Here are couple more benchmarks of the 2TB 5K3000 since there aren't many around:
club.coneco.net said:
5224002l.jpg
5224003l.jpg

====== HddRpmEst v0.1.5 === Report ============================
Target HDD  : Hitachi HDS5C3020ALA
Rotational Speed (estimated) : 5938 rpm
Average Access Time  : 20.1 msec
Average Access Time (10% Short Stroke)  : 11.1 msec
Transfer Rate(Outer,Maximum)  : sustained 149.1 MB/s / burst 160.6 MB/s = 92.8 %
Transfer Rate(Outer,Average)  : sustained 143.3 MB/s / burst 154.7 MB/s = 92.6 %
Transfer Rate(Inner,Average)  : sustained 67.8 MB/s / burst 73.3 MB/s = 92.5 %
Transfer Rate(Inner,Minimum)  : sustained 65.3 MB/s / burst 70.6 MB/s = 92.5 %
Inner/Outer Ratio(Average,Smallest) :  47.3 %, 43.8 %  /    47.4 %, 44.0 %
Remarks         : Outside[29/29,v0.16,a0.94] Inside[24/24,v0.13,a0.96]
------------------------------------------------------------------------------

bbs.kakaku.com said:
836697m.jpg

2TB 5K3000 (P8P67 Deluxe, Marvell Controller)
 
Last edited:
I'm looking at the same 3 units today for data storage. So I want to avoid firmware updates which nixs the F4. Are there concerns with the Hitachi 2TB 5K3000 running on WIN 7 64
 
The 5K3000 drives are relatively new, so there is not much real-world information available about them. If possible, I'd wait a few months to see if any large-scale problems show up with them. People have been banging on the HD204UI drives for months now, and of course the firmware bug turned up, but there is a fix for that, and I am not aware of any other large-scale problems with the HD204UI.
 
There are a few examples of oddities with the SMART data in the F4 thread. The hitachi drive is clearly faster. One thing I do not like about hitachi drives is these do not use 4K sectors. I know loads of other people buy them because of that but I would prefer a lower theoretical URE rate that comes with 4K sectors versus 512 byte ones.
 
but I would prefer a lower theoretical URE rate that comes with 4K sectors versus 512 byte ones.

Except there is no evidence that the BER is lower for 4KB sector drives. The BER specifications are the same.

It is incorrect to say that the actual drives have a lower "theoretical" BER. While it is in theory possible to implement an ECC on 4KB sector HDDs that has lower BER than that on a 512B sector HDD, there is no evidence that any of the manufacturers have done so. Changing the ECC would be a significant hardware and firmware change, and there is no evidence that the HDD manufacturers have YET switched to the type of ECC that would be required to improve the BER.

Unless you have evidence that the manufacturers did not simply use an ECC for the 4KB sector drives that is equivalent to that for the 512B sector drives?
 
Last edited:
What if you had a ZFS pool of samsung F4s using ASHIFT=12, and you add these Hitachis to the pool - will it still work?
 
Just waiting for it to arrive.

The way I figure it, the failure rates can't be any worse than what I have already read about the Samsung, WD, or Seagate 2TB drives...

Its a low performance, high capacity drive, so I will be using it for media storage. From what I have read coolspin or whatever makes it a bit faster than your typical 5400 rpm drive. Sounds like it is variable, but pretty much hits the 5900 that some of the others do. I figured for the price it isn't going to get much cheaper than that anytime soon. Likely the 3TB drives will be expensive for a little bit until there is a lot on the market. So I just hope it is OK and not DOA really...
 
I have 6x 5K3000s in a RAID6 array on my Areca 1880ix and they are working great.
 
I have 23 F4EG drives now (20 in use now and had to buy 3 extra to migrate some data), and - besides the small hassle of upgrading the firmware (which may not be needed with new drives now and just takes a few minutes) - they are great.

I actually used 3 of them in a raidz yesterday to migrate some data (moving from a 7 disk raidz2 to a 10 disk raidz2) and even though that configuration is not be the best capacity wise, I got a consistent 200MB/s write and 300MB/s read - just what you would expect.

I would get the cheapest disks and make sure you have a sound raid setup instead. In my case, using raidz2 (two parity disks) on each set of 10 total disks decreases my chances of a data failure to (IMHO better than) normal enterprise levels with only 25% higher cost. I would rather have 10 cheap disks in raidz2 than "better/faster" drives in a less failsafe configuration.

My most important data (a small subset of my total storage need) is mirrored on both my raidz2s - so at that point at least 6 drives (30%) must fail completely to loose any of my most important data - and even if 12 drives fail at once I could be lucky...
 
Back
Top