une

After much research I still cannot visualize some aspects of hard disk operation. Here are my questions;

1) FORMATTING
When a disk is formatted with a particular file system is the entire surface of the disk changed in some way, or is information deposited in a control portion of the disk. Here are two analogies using a football field as the disk's surface to help describe my understanding;

Analogy a)
Lets say we want to format the football field in NTFS. The greenkeeper goes out and mows the grass in concentric circles. Let's now say we want to reformat it as ext3. The greenkeeper goes out and mows the grass in a criss-cross pattern.

Analogy b)
Lets say we want to format the football field in NTFS. The greenkeeper goes to his office and makes a note in his ground use register that only teams who use NTFS equipment can play on it. Let's now say we want to reformat it as ext3. The greenkeeper goes back to his register and alters the entry to say that only teams who use ext3 equipment can play on it. The grass is mowed any way the lawn mower feels like on the day, it makes no difference.

Which analogy is more correct?

2) PARTITIONING
When a disk is partitioned do the file systems of each partition have to be decided upon when partitioning? What exactly defines a partition? Is a partition defined by what we could call a "line"? Or is it defined as a series of addresses held in a control sector of the disk? Let's go back to the football field analogy.

Analogy c)
The greenkeeper wants to split the football field into 3 fields, 2 half size soccer fields (2 x NTFS) and a half size hockey field (1 x ext3). So he draws 2 permanent dividing lines on the grass. At this point the field is divided (partitioned) but their use has not yet been designated (not formatted yet). Can a disk be partitioned without formatting the partitions?

Analogy d)
The greenkeeper wants to split the football field into 3 fields, 2 half size soccer fields (2 x NTFS) and a half size hockey field (1 x ext3). So he mows 2 areas concentrically (for soccer = NTFS) and mows one in criss cross (for hockey = ext3). Then he makes notes in his register regarding which area is for which sport . There are no "lines" on the grass to be seen.

Which analogy is more correct? Is the partition defined by a "line" between the separate sub-fields, or is it defined by the state of the surfaces of each sub-section of field, or is it defined by information held in a central "control sector", or is the partition defined via a combination of these three aspects?

3) FILE SYSTEMS
How the different file systems work has me puzzled. For example on a disk containing 2 operating systems (say Linux and Windows) the operating systems can see the partitions, but cannot recognize a foreign file system. This leads to to think that a partition consists of a figurative "line" marked on the disk. Or are the partitions defined by information stored in a central disk management sector that Windows for example would read and see "from sector aaa000 to sector adr234 is NTFS, sectors adr235 to zzz999 are foreign to me but I know they are in a separate partition." Linux would read the same disk management sector and see "from sector aaa000 to sector adr234 is something I cannot recognize but I know they are in a partition of their own, sectors adr235 to zzz999 are ext3 and I can use these." Thus an operating system cannot read a foreign file system, but partitions are OS independent. Is this correct?

Say I create a few NTFS partitions on my HD using Windows. I examine the same HD with Linux. Linux can see it has been partitioned but cannot read any data because the file system is foreign. Can Linux then reformat one of these partitions to ext3? If so would this change the partitioning definitions?

Good luck. I have yet to find anyone who can explain in plain English how the three concepts of partitions, file systems and formatting all hang together.

linderman

formatting is a rather complex ordeal and not sure sure if all needs to be digested but here are a few tid bits


partitioning is not file system dependant; partitioning is executed without regard for which file system will be used

the partition setup is nothing more than a map of where the "chunk" starts and where it ends

formatting does not prepare or write to the whole disk space. it too is a map advising the OS and bios how to store bytes of info in 4kb clusters (ntfs) or in larger clusters (fat32)

sorry i cant be of more help than this maybe someone else can ?

une

Thanks, that's a good start. It seems that perhaps the surface of the disk is not physically altered when partitioned or formatted. All these processes do is change some information in a special "control area" of the disk (MBR and/or Partition table?).
I reformatted a 250GB external HD the other day in NTFS. This took a very long time. That's what made me think formatting actually visits each and every "sector(?)" on the HD in the partition being formatted and physically changes it in some way.

linderman

ahhhhhhhh now we are back in familiar water


I gurantee you the disk is not changed/ altered in any way during partitioning or formatting

both partitioning and formatting both are nothing more than a written code for the rules of disk use ( a road map so to speak)

they are nothing more than directions to the system of where to start & stop writing on the disk platters according to cylinders and disks.

you could google research "hard drive geometery" if you want a Tylenol moment

many times a user with a botched or crashed OS will reformat the drive only to find install problems or stability issues. These are the result of the previous install left data bits not wiped during the formatting process'

when this occurs a simple zero fill utility is used (like boot n nuke) to write zeros during several passes to each sector of the hard drive, now this is SLOWWWWWWWW it takes about one hour to zero fill 20gigs of drive space!

the zero fill also removes all previous partitions and formatting info/ file system blah blah and gives you the same fresh start you have with a brand new drive out of the box.

enjoy

une

Thanks again. So would it be true to say that partitioning is totally OS independant? I think it is, but the following creates some doubts in my mind about this; when you create a partition you are asked what file system the partition will use, so it seems that partitioning and file systems are tied in together somehow.

Can I create an NTFS partition on my HD using Windows, then using Linux reformat this partition to ext3? I am thinking that perhaps the partitioning info (start and stop sectors on the HD) remains constant and all OSs read this part of the "HD road map" in the same way. However another piece of data in the "HD road map" is file system type. This can be changed at will without disturbing the partition boundaries. Leave the partition boundary settings in the "HD road map" alone, change the file system type entry in the "HD road map" and voila, you have reformatted the partition. Changing the file system type of the partition in this way is actually synonomous with reformatting the partition and will render the data in that partition unreadable, unreachable and useless. Only new data written by the OS which reformatted the partition will then be usable. Do I have all this correct?

linderman

yes you have it correct; just remember linux doesnt use NTFS

I am not sure how you would reformat a partition once it has been NTFS to tansform it to Fat32 (linux can use fat32)

you can easily convert from fat32 to NTFS but not vica versa without zero filling the drive

the best way IMHO to set-up your endeavor is to have linux on its own drive and the windows OS on its own drive; then you simply boot from which ever one you desire at the moment