What's the real physical geometry of WD20EARS? Should I care?


#1

Hello.

Solving the problem with correctly aligning partitions on my Linux server, I came across the question about real physical geometry of this drive. I can’t find this info. Or, may be it is irrelevant?

I just used the following scheme for partitioning: start partition at mod8 sector number (I used parted in “unit s” mode) and end partition at mod8 - 1 sector number (to start at mod8 next one). But now I think, that, may be, I should consider starting new partition not only on new big sector (4k sector), but also at cylinder boundary?

But… this drive reports to OS invalid CHS data (at least it lies about sector size), so… How can I get it’s real geometry? Or I shouldn’t care about it so much and aligning just to 4k sector is ok?


#2

Since there’s been no replies, maybe you should try contacting WD’s Technical Support about this. You can do so either by phone or email.

To Contact WD for Technical Support
http://support.wdc.com/contact/index.asp?lang=en


#3

Your question is ambiguous.

For the past 20 years, ever since address translation (and subsequently Zone Bit Recording) was introduced for IDE drives, the CHS geometry reported by a drive has always been “fake”. That is to say, no drive has 255 heads, or a constant number of sectors per track across the whole surface. Nowadays, with the introduction of VBPI (Variable Bits Per Inch) and VTPI (Variable Tracks Per Inch), even two drives within the same model may not have the same physical layout. In fact each drive has a translator module in the firmware to handle the mapping between LBAs and real Cylinders, Heads, and Sectors.

In any case, I believe modern partitioning schemes do not care whether partitions are aligned to cylinder boundaries. However, some partitioning software may issue harmless warnings if such conditions are detected.