HardwareBanter

HardwareBanter (http://www.hardwarebanter.com/index.php)
-   General (http://www.hardwarebanter.com/forumdisplay.php?f=28)
-   -   ST4000NM0035 vs ST4000NM0033 (http://www.hardwarebanter.com/showthread.php?t=196905)

Lars Bonnesen December 18th 16 01:32 PM

ST4000NM0035 vs ST4000NM0033
 
For at 4 TB drive sitting internally in a SuperMicro server as an
ESX-test-/develserver, would you prefer the newer Seagate ST4000NM0035 over
ST4000NM0033?

I can see that the newer one are having a slightly faster theoretically
internal speed, but otherwise they look quite the same. Are the newer ones
worth the price difference?

Regards, Lars.


Paul[_28_] December 18th 16 02:40 PM

ST4000NM0035 vs ST4000NM0033
 
Lars Bonnesen wrote:
For at 4 TB drive sitting internally in a SuperMicro server as an
ESX-test-/develserver, would you prefer the newer Seagate ST4000NM0035
over ST4000NM0033?

I can see that the newer one are having a slightly faster theoretically
internal speed, but otherwise they look quite the same. Are the newer
ones worth the price difference?

Regards, Lars.


ST4000NM0033 is 5 platter, 10 head. 175MB/sec OD
ST4000NM0035 is 4 platter, 7 head. 205MB/sec OD --- higher areal density

http://www.seagate.com/www-content/p...100671511f.pdf
http://www.seagate.com/www-content/p...100793636c.pdf

The '35 is 1 inch high.
The '33 is 1 inch high.

Both are 512n.

The 7 head drive will have 8 heads, but one head
will be ignored. Having two heads on each platter
balances the forces on the spindle (to support all
compass points as mounting options).

StorageReview has pictures of the 33, but nothing on 35 that I could find.

http://www.storagereview.com/seagate...n_e s3_review

'33 has some failures. One here at 3 months.

http://www.newegg.com/Product/Produc...82E16822178307

'35 has no reviews yet by customers.

http://www.newegg.com/Product/Produc...1Z4-002P-00141

The newer one should actually be cheaper, if all that
mattered was platter and head count. But you know there is
more to it than that. Both drives mysteriously weigh the
same amount. So the '35 has more metal in the chassis... or
something.

Paul

Lars Bonnesen December 18th 16 06:04 PM

ST4000NM0035 vs ST4000NM0033
 
Yeah - the density is higher, so the internal speed is therefor also higher.

1 platter less, lighter and cheaper, but maybe it has better bearings

Actually I would imagine that one platter less should make it slower...

Maybe I should stick to the older one?

Regards, Lars.

"Paul" wrote in message ...

Lars Bonnesen wrote:
For at 4 TB drive sitting internally in a SuperMicro server as an
ESX-test-/develserver, would you prefer the newer Seagate ST4000NM0035
over ST4000NM0033?

I can see that the newer one are having a slightly faster theoretically
internal speed, but otherwise they look quite the same. Are the newer ones
worth the price difference?

Regards, Lars.


ST4000NM0033 is 5 platter, 10 head. 175MB/sec OD
ST4000NM0035 is 4 platter, 7 head. 205MB/sec OD --- higher areal
density

http://www.seagate.com/www-content/p...100671511f.pdf
http://www.seagate.com/www-content/p...100793636c.pdf

The '35 is 1 inch high.
The '33 is 1 inch high.

Both are 512n.

The 7 head drive will have 8 heads, but one head
will be ignored. Having two heads on each platter
balances the forces on the spindle (to support all
compass points as mounting options).

StorageReview has pictures of the 33, but nothing on 35 that I could find.

http://www.storagereview.com/seagate...n_e s3_review

'33 has some failures. One here at 3 months.

http://www.newegg.com/Product/Produc...82E16822178307

'35 has no reviews yet by customers.

http://www.newegg.com/Product/Produc...1Z4-002P-00141

The newer one should actually be cheaper, if all that
mattered was platter and head count. But you know there is
more to it than that. Both drives mysteriously weigh the
same amount. So the '35 has more metal in the chassis... or
something.

Paul


Lars Bonnesen December 18th 16 06:06 PM

ST4000NM0035 vs ST4000NM0033
 
Reliabillity is supposed to be better on the newer one:

https://www.span.com/compare/ST4000N...33/58710-39360

Regards, Lars.


Paul[_28_] December 18th 16 07:07 PM

ST4000NM0035 vs ST4000NM0033
 
Lars Bonnesen wrote:
Reliabillity is supposed to be better on the newer one:

https://www.span.com/compare/ST4000N...33/58710-39360

Regards, Lars.


Roughly the ratio of five platters to four platters,
heads etc. And those are MTBF calculations, not actual
field data. If Seagate gave us actual field data collected
from previous models, I think you would be shocked at the
correlation between MTBF and actual.

Paul

Lars Bonnesen December 19th 16 07:28 AM

ST4000NM0035 vs ST4000NM0033
 
But with a better MTBF, I guess Seagate expect the newer ones to be more
reliable than the older ones.

Regards, Lars.

"Paul" wrote in message ...

Lars Bonnesen wrote:
Reliabillity is supposed to be better on the newer one:

https://www.span.com/compare/ST4000N...33/58710-39360

Regards, Lars.


Roughly the ratio of five platters to four platters,
heads etc. And those are MTBF calculations, not actual
field data. If Seagate gave us actual field data collected
from previous models, I think you would be shocked at the
correlation between MTBF and actual.

Paul


Paul[_28_] December 19th 16 08:51 AM

ST4000NM0035 vs ST4000NM0033
 
Lars Bonnesen wrote:
But with a better MTBF, I guess Seagate expect the newer ones to be more
reliable than the older ones.

Regards, Lars.


That's what the measure is intended to communicate.

Normally, you would use the MTBF as a prediction of
how many spares to stock in your stock room. I doubt
the companies running large server farms rely on
that manufacturer number, instead looking at how many failed
Seagates or WDCs they had when using the previous
model.

MTBF is more of a ceremonial dance, than a useful metric.
The manufacturer does have the information necessary to
make the number realistic. However, if they did that,
the marketing department would have the individuals
killed :-) So instead, the calculation is done in
"a standard way", devoid of the most recent field
data. That's why I object to it. There is really nothing
forcing them to be completely honest.

As long as the numbers stay in a small range, nobody
will suspect anything is wrong with the numbers.

As an example, power supplies at work, the bigger
ones are rated at 3000 FITs. And then a lab prototype
of a smaller converter, had a rating of 100 FITS. And
when someone calculates a value like that, you immediately
start grilling them with questions. But if you showed me
one unit is 3000, another 2500, then my suspicions are
not aroused. Like you, I might expect (magically)
that somehow the 2500 FIT one is better. (FIT equals
failure in 10^9 hours). When there is an extreme change
in these reliability numbers is when you become suspicious.

To give another example of dishonesty, you will notice that
some Helium drives have entered the market. And I was hoping
that the environmental spec would take a radical shift.
Instead, the specs are almost verbatim copied from
existing drives. And you know when that happens,
the marketing department said "don't scare the customers,
don't arouse suspicions". The max altitude on a sealed
Helium drive should be higher than the previous drives.
Or at least, the number should be different, not
exactly the same by chance.

I'm patiently waiting for some failure rate data
to come back from the Helium drives. By being
sealed, they should not be subject to external
moisture effects on the platters. Now, I wonder
what the MTBF figures say... Would they reflect
a difference ? Or not ?

Paul

[email protected] March 13th 19 03:02 PM

ST4000NM0035 vs ST4000NM0033
 
On Sunday, 18 December 2016 18:02:41 UTC+5:30, Lars Bonnesen wrote:
For at 4 TB drive sitting internally in a SuperMicro server as an
ESX-test-/develserver, would you prefer the newer Seagate ST4000NM0035 over
ST4000NM0033?

I can see that the newer one are having a slightly faster theoretically
internal speed, but otherwise they look quite the same. Are the newer ones
worth the price difference?

Regards, Lars.


Can we use both ST4000NM0033 and ST4000NM0035 in a RAID6, i.e for example can we use 2 nos. ST4000NM0033 and 2 nos ST4000NM0035 ??

Paul[_28_] March 13th 19 06:18 PM

ST4000NM0035 vs ST4000NM0033
 
wrote:
On Sunday, 18 December 2016 18:02:41 UTC+5:30, Lars Bonnesen wrote:
For at 4 TB drive sitting internally in a SuperMicro server as an
ESX-test-/develserver, would you prefer the newer Seagate ST4000NM0035 over
ST4000NM0033?

I can see that the newer one are having a slightly faster theoretically
internal speed, but otherwise they look quite the same. Are the newer ones
worth the price difference?

Regards, Lars.


Can we use both ST4000NM0033 and ST4000NM0035 in a RAID6, i.e
for example can we use 2 nos. ST4000NM0033 and 2 nos ST4000NM0035 ??


For RAID, it's preferable to have drives with TLER (time limited error recovery).
That helps prevent unnecessary array rebuilds due to the reallocation of
sectors.

TLER/CCTL

https://en.wikipedia.org/wiki/Error_recovery_control

Other than that, you could likely mix those two drives in RAID,
as the slowest characteristic determines the overall performance.
If one drive does 200MB/sec and the second drive does 210MB/sec,
the 200MB/sec drive defines the response. If one has a seek
time of 12ms and another drive has a seek time of 13ms, the
13ms drive determines the response of the array (for anything
which cannot be hidden by that particular RAID array type).

You also select drives rated for the vibration characteristic
of the application. If the disks are going in a rack mount
with a lot of other drives, an Enterprise drive might be
better for that than a Consumer drive. An Enterprise drive
with a piezo actuator at the head for fine positioning,
is supposed to track better when the drive receives vibration
from the drives next to it.

Those are all considerations when selecting drives for RAID.

Paul


All times are GMT +1. The time now is 04:50 PM.

Powered by vBulletin® Version 3.6.4
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
HardwareBanter.com