Modern HDs and Zone Bit Recording

[Franc Zabkar]

Here is what HD Tune's developer(s) consider to be a typical
performance graph for a hard drive (160GB Maxtor):

http://www.hdtune.com/HDTune_Benchmark.gif

The graph shows 16 steps. AIUI, this corresponds to 16 zones of
constant angular bit density and is referred to as Zone Bit Recording.

Here are the performance graphs for two Seagate ST31000520AS drives:

http://i32.tinypic.com/2saargj.jpg

http://i29.tinypic.com/5y6hyf.jpg

The first has no obvious zones, whereas the second has 40
discontinuities. Are these discontinuities ZBR related?

Do modern drives still use ZBR, or does the linear bit density remain
constant across all cylinders?

- Franc Zabkar

 

[2312]

Here is what HD Tune's developer(s) consider to be a typical
performance graph for a hard drive (160GB Maxtor):

http://www.hdtune.com/HDTune_Benchmark.gif

The graph shows 16 steps. AIUI, this corresponds to 16 zones of
constant angular bit density and is referred to as Zone Bit Recording.

Here are the performance graphs for two Seagate ST31000520AS drives:

http://i32.tinypic.com/2saargj.jpg

http://i29.tinypic.com/5y6hyf.jpg

The first has no obvious zones, whereas the second has 40
discontinuities. Are these discontinuities ZBR related?

Do modern drives still use ZBR,

Yep, and you can see that in the datasheets of the drives that have detailled datasheets.

or does the linear bit density remain constant across all cylinders?

Nope.

 

[Arno]

Here is what HD Tune's developer(s) consider to be a typical
performance graph for a hard drive (160GB Maxtor):

The graph shows 16 steps. AIUI, this corresponds to 16 zones of
constant angular bit density and is referred to as Zone Bit Recording.

Here are the performance graphs for two Seagate ST31000520AS drives:


The first has no obvious zones, whereas the second has 40
discontinuities. Are these discontinuities ZBR related?

Do modern drives still use ZBR, or does the linear bit density remain
constant across all cylinders?

ZBR, the track lengths are just too different to throw away all
those extra bits.

Arno

 

[Franc Zabkar]

ZBR, the track lengths are just too different to throw away all
those extra bits.

Arno

By "constant linear bit density" I was suggesting that bits per inch
could remain the same for every track, ie I was asking whether each
track is effectively a new zone.

- Franc Zabkar

 

[Franc Zabkar]

I didn't have an ST3100520AS. I did have an ST31500341AS, which I
think is in the same series.

1. The wild fluctuations suggest something is going on that is messing
up the results.
2. If ones computer can't handle the full sequential speed of the
disk then the first part of the graph will be flat. (I don't
think that is the problem on your system, but I mention it for
completeness.
3. What Options (click on the "gears" icon) are you using for
"Benchmark"? The are two variables: "Test speed/accuracy"
and "Block size". I don't know exactly what each does,
but your goal is the fastest disk speed ("Block size" affects
this) and best accuracy (I think "Fast" versus "Accurate"
correctly indicates the tradeoffs)

The subject HDs are not mine. I was watching the following thread in
Seagate's forums:

http://forums.seagate.com/stx/board/message?board.id=ata_drives&message.id=14942

I just Read Benchmark for my Seagate ST31500341AS in my Dell Precision
380 running Window XP Professional with Service Pack 3 using
HD Tune Pro V3.50 (corresponds to non-pro V2.55, I think). I used
the most accurate setting and blocksize 128KB.

The speed went held at about 125MB/s until 10% then went down
to about 60MB/s at 100%, looking like a straight line with "wiggles".
There were no wild swings, but the "wiggles" in the data points were
perhaps +/- 2MB/s, which is more than I would expect the difference
between zones would be, given that there might be 40 or so zones. (In
other words, the plot looks more like a straight line with wiggles
than a staircase with wiggles.) (Note that either my system isn't
able to read the disk a full speed or the first zone is bigger than
the rest of the zones.)

Using Block size 8MB seemed to increase the wiggle size but decrease
the frequency of the wiggles. I.e., still seems more like a straight
line after 10% than a staircase.

Seagate's page:
http://www.seagate.com/ww/v/index.jsp?name=st31000520as-bcuda-lp-sata-1tb-hd&
vgnextoid=9d373f15020b0210VgnVCM1000001a48090aRCRD&vgnextchannel=

f424072516d8c010VgnVCM100000dd04090aRCRD&locale=en-US&reqPage=Support>
says the ST31000520AS is
Barracuda Low-Power SATA 3Gb/s 1TB Hard Drive
with capacity 1TB, cache 32MB, 5900RPM

Seagate's page:
http://www.seagate.com/ww/v/index.jsp?name=st31500541as-bcuda-lp-sata-1.5tb-hd&
vgnextoid=91abe5daa90b0210VgnVCM1000001a48090aRCRD&vgnextchannel=
f424072516d8c010VgnVCM100000dd04090aRCRD&locale=en-US&reqPage=Support
says the ST31500541AS is
Barracuda Low-Power SATA 3Gb/s 1.5TB Hard Drive
with capacity 1.5TB, cache 32MB, 5900RPM


The "Product Manual" for the "Barracuda LP Series SATA" including
ST230000542AS, ST31500541AS and ST3100520AS is in:
http://www.seagate.com/staticfiles/support/disc/manuals/desktop/Barracuda LP/100564361b.pdf

I didn't see mention of zones or the track lays.

I've checked a few product manuals also.

Also, at least two
places (page 6 and page 8) said the "Sustained data transfer rate OD"
was "95 Mbytes/sec max", but I see faster reads on big files (I
haven't measured the write rates.)

Thanks for your input.

Here is another Seagate forum thread:
http://forums.seagate.com/stx/board/message?board.id=ata_drives&thread.id=13875

This graph appears to show about 20 zones:
http://img361.imageshack.us/img361/5630/hdtunebenchmarkst310005.png

There are discontinuities, but not all appear to occur at the zone
boundaries.

- Franc Zabkar

 

[Rod Speed]

By "constant linear bit density" I was suggesting that bits per inch
could remain the same for every track, ie I was asking whether each
track is effectively a new zone.

Trouble with that approach is that needs too big a table
to convert logical block numbers to track numbers.

 

[Rod Speed]

Trouble with that approach is that needs too big a table
to convert logical block numbers to track numbers.

And would be an absolute dog to implement with the initial formatting.

 

[Arno]

On 10 Sep 2009 10:26:02 GMT, Arno <[email protected]> put finger to
keyboard and composed:

By "constant linear bit density" I was suggesting that bits per inch
could remain the same for every track, ie I was asking whether each
track is effectively a new zone.

They use ZBR. Reason is likely simplicity.

Arno