Actual data transfer rates between drives

[Valmiki]

Is there a way to measure the actual data transfer rate between separate
physical drives in a computer?

I figured that with a couple of SATA drives (150MB/sec) the rate would be
about 8 GB a minute and it will take roughly 30 minutes to transfer the
entire contents of one filled 250 GB unit to a blank 500 GB drive. The 250
GB unit is the secondary data drive being replaced with the larger one.

The task actually took just under 3 hours. This NForce3 motherboard-based
machine works perfectly well and is still blazingly fast otherwise.

No complaints here, but I was just wondering.

TIA

Valmiki
--------

 

[Paul]

Is there a way to measure the actual data transfer rate between separate
physical drives in a computer?

I figured that with a couple of SATA drives (150MB/sec) the rate would be
about 8 GB a minute and it will take roughly 30 minutes to transfer the
entire contents of one filled 250 GB unit to a blank 500 GB drive. The 250
GB unit is the secondary data drive being replaced with the larger one.

The task actually took just under 3 hours. This NForce3 motherboard-based
machine works perfectly well and is still blazingly fast otherwise.

No complaints here, but I was just wondering.

TIA

Valmiki
--------

There is a difference between the cable transfer rate (150MB/sec) and
the sustained media rate where the head meets the platter.

On my disk, the sustained transfer rate, at the beginning of my disk,
is 60MB/sec. Near the end of the disk, the transfer rate drops to
40MB/sec.

If you benchmark with HDTach ot HDTune, you will be able to see this.

The sustained transfer rates, can also be seen in the Storagereview
database, here. The maximum transfer rate of the WD Caviar SE16, at
97MB/sec, is more than 50% better than my drive. SCSI drives, the
ones that rotate at 15K RPM, have the best results, at 135MB/sec.

http://www.storagereview.com/Testbed4Compare.sr

When there are short bursts of data, they can flow into the cache memory
chip on the disk drive. But once the cache is full, you are limited to
the "drain rate", as the sectors are transferred to the platters. And
that is when you see those "60MB/sec" type numbers.

This is a transfer curve for a WD Raptor, a 10K SATA drive. The
curve shows the sustainable transfer rate, across the disk surface,
from beginning to end. It is getting about 87MB/sec sustained,
near the beginning of the disk.

http://images.anandtech.com/iblog/raptor150sb600hdtach301S.jpg

Paul

 

[Valmiki]

My response at the bottom:

There is a difference between the cable transfer rate (150MB/sec) and
the sustained media rate where the head meets the platter.

On my disk, the sustained transfer rate, at the beginning of my disk,
is 60MB/sec. Near the end of the disk, the transfer rate drops to
40MB/sec.

If you benchmark with HDTach ot HDTune, you will be able to see this.

The sustained transfer rates, can also be seen in the Storagereview
database, here. The maximum transfer rate of the WD Caviar SE16, at
97MB/sec, is more than 50% better than my drive. SCSI drives, the
ones that rotate at 15K RPM, have the best results, at 135MB/sec.

http://www.storagereview.com/Testbed4Compare.sr

When there are short bursts of data, they can flow into the cache memory
chip on the disk drive. But once the cache is full, you are limited to
the "drain rate", as the sectors are transferred to the platters. And
that is when you see those "60MB/sec" type numbers.

This is a transfer curve for a WD Raptor, a 10K SATA drive. The
curve shows the sustainable transfer rate, across the disk surface,
from beginning to end. It is getting about 87MB/sec sustained,
near the beginning of the disk.

http://images.anandtech.com/iblog/raptor150sb600hdtach301S.jpg

Paul

Thanks for the informative response. It explained matters more clearly than
I thought myself.

I see that my drives actually attained a sustained data transfer rate of
20MB/sec - which is actually not too bad, I think!

Regards and thanks again.

Valmiki
--------

 

[Paul]

My response at the bottom:



Thanks for the informative response. It explained matters more clearly than
I thought myself.

I see that my drives actually attained a sustained data transfer rate of
20MB/sec - which is actually not too bad, I think!

Regards and thanks again.

Valmiki

OK, I should explain something.

First, get a copy of a benchmarking program.

http://www.simplisoftware.com/Public/index.php?request=HdTach

If you look at their sample image, notice that the transfer rate
is a "curve", and diminishes as you go further away from sector zero.

If you run the benchmark (which is a read-only benchmark), and if
you see a "flat line" on the screen, that also tells you something.
It means the transfer is not media limited, and is cable limited
in some way. This happens sometimes on IDE cables, and Windows
can decrease the transfer rate used, based on detecting excessive
CRC errors.

So, give HDTach a try, and see what shape the transfer curve is.

If you are transferring a lot of small files, from one disk to the
other, then there could be more overhead in doing so. If you
are transferring very large files, you should be getting
closer to the sustained transfer rate limit. Running that
benchmark, controls conditions better than using Windows
file copy. So verify that both drives have proper transfer
rate curves first, before concluding everything is working
perfectly.

And if you suspect your transfer rate is being limited,
check this KB article. The "workaround" section, describes
a way to restore full rate transfer.

http://support.microsoft.com/default.aspx?scid=kb;en-us;817472

This transfer is so slow, you'd suspect the interface to the
drive was running in PIO (polled) transfer mode. Performance
could likely be improved, by using the KB817472 article.

http://www.yesky.com/imagesnew/digi/0412/08/t25.gif

Paul

 

[Valmiki]

My response at the bottom:

OK, I should explain something.

First, get a copy of a benchmarking program.

http://www.simplisoftware.com/Public/index.php?request=HdTach

If you look at their sample image, notice that the transfer rate
is a "curve", and diminishes as you go further away from sector zero.

If you run the benchmark (which is a read-only benchmark), and if
you see a "flat line" on the screen, that also tells you something.
It means the transfer is not media limited, and is cable limited
in some way. This happens sometimes on IDE cables, and Windows
can decrease the transfer rate used, based on detecting excessive
CRC errors.

So, give HDTach a try, and see what shape the transfer curve is.

If you are transferring a lot of small files, from one disk to the
other, then there could be more overhead in doing so. If you
are transferring very large files, you should be getting
closer to the sustained transfer rate limit. Running that
benchmark, controls conditions better than using Windows
file copy. So verify that both drives have proper transfer
rate curves first, before concluding everything is working
perfectly.

And if you suspect your transfer rate is being limited,
check this KB article. The "workaround" section, describes
a way to restore full rate transfer.

http://support.microsoft.com/default.aspx?scid=kb;en-us;817472

This transfer is so slow, you'd suspect the interface to the
drive was running in PIO (polled) transfer mode. Performance
could likely be improved, by using the KB817472 article.

http://www.yesky.com/imagesnew/digi/0412/08/t25.gif

Paul

Great. Very informative. Have downloaded HDTach and will give it a spin
sometime soon!

Regards and thanks.

Valmiki
--------