$160 10TB WD Easystore External HDD back at Best Buy

ac12

Mu-43 All-Pro
Joined
Apr 24, 2018
Messages
1,517
Location
SF Bay Area, California, USA
Always crazy when externals are cheaper than internal HDD.
The devil is in the details.
The drive in that external case is probably a cheaper, high capacity but low performance (lower RPM) drive.
IOW, not level of drive that I would put in my computer.
But for pure data storage, not speed of data transfer, that is OK.
 

tkbslc

Super Moderator
Joined
Feb 6, 2015
Messages
7,413
Location
Salt Lake City, UT, USA
The devil is in the details.
The drive in that external case is probably a cheaper, high capacity but low performance (lower RPM) drive.

I found someone that pulled the drive and tested it and it's a great performer:

WD WD100EMAZ and WD Easystore 10TB External Backup Drive Review - Page 3 of 4

They did say they had to mask a pin on the sata power cable to get it to work in some devices.

But even if it was a "slow" HDD, That seems irrelevant when the cheapest internal 5400RPM 10TB drive that WD makes is $270. That's 69% more expensive than this external drive.
 

ac12

Mu-43 All-Pro
Joined
Apr 24, 2018
Messages
1,517
Location
SF Bay Area, California, USA
The AIDA64 Linear Write test is what is most interesting.
The reason is, once you fill the buffer, you hit the bottleneck of head to disk data transfer, or sustained data transfer rate.
The sustained data transfer rate is some/many times not in the consumer specs, because it is so much lower than the interface rate.
When I am backing up Gigabytes of data, I FILL the butter in a minute, so it is sustained data transfer rate that is the choke point.

All things being equal (drive capacity, number of platters, number of heads, number of tracks, etc.), a 7000 (7k) RPM drive will have a faster sustained data transfer rate than a 5000 (5k) RPM drive.

When your backup takes more than 10 hours to run, this makes a difference. And this would not be unusual with huge photo and video libraries. If 100GB takes an hour to transfer via USB3, 1TB will take about 10 hours.

I am current running a full backup. One month of photos takes about 1 hour to backup from a 7k RPM data drive to an external (probably 5k RPM) drive. So I am looking at the entire backup taking about 2+ days to do.
One of these days, I am going to make a 7k external drive, just to see how much faster it will go.​

USB 3 will do 600MB/s data xfer, so that is not a limiting factor, the drives on either end are.
Ideally you have a 7k RPM data drive in your computer, so it isn't the choke point, and a 7k RPM backup drive.
If you have a 5k RPM data drive, using a 7k RPM backup drive does not make sense, because the data drive cannot send the data fast enough to keep up with the backup drive. So the data drive becomes the choke point.

So like many things it comes down to speed and time vs. cost.
Lower cost = slower speed and longer backup time.
Higher cost = faster speed and shorter backup time.
 

masayoshi

Mu-43 Top Veteran
Joined
Dec 5, 2016
Messages
902
Location
Salt Lake City
Real Name
Masaaki
This is a great drive if you do the incremental backup.
Sure it will take long time to do full back up, but if you use software that copies and backup only the additions/changes, then it will be a matter of minutes even if you have several terabytes of data. Of course if you add terabytes of data every week, then it's a different story.
 

tkbslc

Super Moderator
Joined
Feb 6, 2015
Messages
7,413
Location
Salt Lake City, UT, USA
USB 3 will do 600MB/s data xfer, so that is not a limiting factor, the drives on either end are.
Ideally you have a 7k RPM data drive in your computer, so it isn't the choke point, and a 7k RPM backup drive.
If you have a 5k RPM data drive, using a 7k RPM backup drive does not make sense, because the data drive cannot send the data fast enough to keep up with the backup drive. So the data drive becomes the choke point.

.
Disk RPM alone does not have a very good correlation to drive sequential transfer speed. Platter density is more of a factor than rotational speed. For example, the 5400RPM 10TB WD Red will do almost 220MB/s, which is as good as the newest 7200RPM 6TB WD Black and almost 50% faster than the previous model 4TB Black (150MB/s). If you still have a 1-2TB 7200RPM drive, it will undoubtedly be much slower than a newer high capacity 5400RPM drive. This is also true in the server world where the newer high density 2.5" 10K drives (and even some 7200RPM ones) are faster than the smaller, lower density 15K drives they replaced.

Higher RPM benefit is primarily for seek time, which affects Random I/O. However ANY hard drive, even 15K, is so bad at random I/O compared to an SSD that it's irrelevant. It's like asking which snail is the slowest. You just wouldn't use a single HDD when you need Random I/O.

Main point is you should check the drive and its real world performance instead of just going by RPM.
 

ac12

Mu-43 All-Pro
Joined
Apr 24, 2018
Messages
1,517
Location
SF Bay Area, California, USA
Disk RPM alone does not have a very good correlation to drive sequential transfer speed. Platter density is more of a factor than rotational speed. For example, the 5400RPM 10TB WD Red will do almost 220MB/s, which is as good as the newest 7200RPM 6TB WD Black and almost 50% faster than the previous model 4TB Black (150MB/s). If you still have a 1-2TB 7200RPM drive, it will undoubtedly be much slower than a newer high capacity 5400RPM drive. This is also true in the server world where the newer high density 2.5" 10K drives (and even some 7200RPM ones) are faster than the smaller, lower density 15K drives they replaced.

Higher RPM benefit is primarily for seek time, which affects Random I/O. However ANY hard drive, even 15K, is so bad at random I/O compared to an SSD that it's irrelevant. It's like asking which snail is the slowest. You just wouldn't use a single HDD when you need Random I/O.

Main point is you should check the drive and its real world performance instead of just going by RPM.
Agree

Data density is another variable.
That was sort-of implied when you work the formula of drive capacity / number of data heads / number of tracks.
Although data per track varies, as the disk companies usually put tracks into zones, with more data sectors on the outer track zones.

I did say
All things being equal (drive capacity, number of platters, number of heads, number of tracks, etc.), a 7000 (7k) RPM drive will have a faster sustained data transfer rate than a 5000 (5k) RPM drive.
Tangent: About I/O, in an extreme example of what you said. Several years back SAP went to replicating parts of the database into muti-hundred GB to multi-TB RAM for that exact reason, to eliminate the disc I/O. Rotating disk I/O had become a serious performance drag for reporting and data analysis. I do not know the exact architecture of the servers.
 

tkbslc

Super Moderator
Joined
Feb 6, 2015
Messages
7,413
Location
Salt Lake City, UT, USA
Well we are probably not that far in disagreement. I just don't think you say all 7200 RPM drives are faster and 5400 slower, that's all. This 5400 RPM drive in the related deal seems to be actually quite fast for file copy.

With regards to your tangent, for the high end databases I design systems for we use both large amounts of RAM and SSD. The entire DB is cached in RAM for reads and the SSD for writes. Prior to SSD, we had to use arrays of 100+ disks and caching controllers to get the IOPS reasonable. But, Non-Volatile RAM will be hitting the mainstream in the next few years and then we can read and write critical data to RAM, as there won't be any power loss concerns. The RAM will not lose data when power is lost! Then we can go back to slow drives for the back end as the work will all happen in memory.
 

ac12

Mu-43 All-Pro
Joined
Apr 24, 2018
Messages
1,517
Location
SF Bay Area, California, USA
ARGH
As I sit here watching a backup of a photo directory at a blistering 5MB/sec.
I think the file system overhead for copying all those small jpg files is killing the data transfer rate.
 

tkbslc

Super Moderator
Joined
Feb 6, 2015
Messages
7,413
Location
Salt Lake City, UT, USA
ARGH
As I sit here watching a backup of a photo directory at a blistering 5MB/sec.
I think the file system overhead for copying all those small jpg files is killing the data transfer rate.
If it is a straight file copy type job, then there is something wrong. That would be a sequential copy and should be full drive speed. You could try defragmenting, but even a a fragmented drive shouldn't be running that slow.

If it is software that creates backup files and you are doing an incremental type backup of only the changes, then it can be slow. Comparing the backup file to the existing data involves a lot of random I/O which is miserable on HDD. Random I/O on a HDD probably would be as slow as 5MB/s.
 
Joined
Mar 17, 2011
Messages
321
Location
USA, Northeast Coast
Real Name
Richard Correale
A good deal for sure but I'm done with hard drives that aren't solid state. I've gone through two over the past 15 years, they just wear out. I also use Carbonite to back everything up but I prefer an additional back up to that also as I store image files for clients.
 

tkbslc

Super Moderator
Joined
Feb 6, 2015
Messages
7,413
Location
Salt Lake City, UT, USA
A good deal for sure but I'm done with hard drives that aren't solid state. I've gone through two over the past 15 years, they just wear out. I also use Carbonite to back everything up but I prefer an additional back up to that also as I store image files for clients.
That works fine for about 2TB worth, but a 10TB SSD is big money.
 
Joined
Dec 31, 2012
Messages
1,101
Well we are probably not that far in disagreement. I just don't think you say all 7200 RPM drives are faster and 5400 slower, that's all. This 5400 RPM drive in the related deal seems to be actually quite fast for file copy.

With regards to your tangent, for the high end databases I design systems for we use both large amounts of RAM and SSD. The entire DB is cached in RAM for reads and the SSD for writes. Prior to SSD, we had to use arrays of 100+ disks and caching controllers to get the IOPS reasonable. But, Non-Volatile RAM will be hitting the mainstream in the next few years and then we can read and write critical data to RAM, as there won't be any power loss concerns. The RAM will not lose data when power is lost! Then we can go back to slow drives for the back end as the work will all happen in memory.

Exactly. RPM is just one of many factors that determines throughput. The interface (USB 2.0, USB 3.0, etc.), a size of the buffer, whether it is a hybrid drive (SSD + drive), etc.
 

cnyap

Mu-43 Regular
Joined
Jan 30, 2015
Messages
172
I’d be more concerned with data longevity but unsure if that spec is published.
 

Latest posts

Top Bottom