The reason why it is good enough to use a tape is because alignment is not critical. It is enough that sectors that is on top of each other with the same number but at different platter is being read at the same time before the timer runs out. This time is usually +- half a sector. A file is not stored in one platter, but is divided into chunks and divided into all the platters. This is for faster transfer. Each sector on each platter got its own preamble and sync word. Thus, there is no need for bit to bit precision. This wouldn't work either because a slightly different temperature on each platter will move a bit far away, many bits from another bit on another platter, in fact, this is the case even for two bits on opposite side of a platter. It was a time though, with mfm encoding (typically disk less than 100 MB) where there were only one synchronization for the whole drive and where all bits on different platter needed to be exactly where it was with a precision better than 10% of a bit. When the bits got smaller this was an impossible encoding scheme and everybody went for rll instead where every physical sector had its own synchronization. With MFM it would not be good enough to use tape. Then you would need a HDD to backup side by side of each platter, re-construct the synchronization and put the data result back on platters in the donor drive. You would also have a challenge to sync the tracks with the head of the new drive because two drives wouldn't have the same track sync. This meant that you would need a low level format where you only ran the part where track sync was set up. In modern hdds such is done in init when a disk boots up and will track, sector- sync continuously and while reading a sector. If data read from the disk falls under a certain threshold the head will be moved until it is on top of the track again, and this without interrupt the read. Write doesn't have direction error control.
The reason why it is good enough to use a tape is because alignment is not critical. It is enough that sectors that is on top of each other with the same number but at different platter is being read at the same time before the timer runs out. This time is usually +- half a sector. A file is not stored in one platter, but is divided into chunks and divided into all the platters. This is for faster transfer. Each sector on each platter got its own preamble and sync word. Thus, there is no need for bit to bit precision. This wouldn't work either because a slightly different temperature on each platter will move a bit far away, many bits from another bit on another platter, in fact, this is the case even for two bits on opposite side of a platter. It was a time though, with mfm encoding (typically disk less than 100 MB) where there were only one synchronization for the whole drive and where all bits on different platter needed to be exactly where it was with a precision better than 10% of a bit. When the bits got smaller this was an impossible encoding scheme and everybody went for rll instead where every physical sector had its own synchronization. With MFM it would not be good enough to use tape. Then you would need a HDD to backup side by side of each platter, re-construct the synchronization and put the data result back on platters in the donor drive. You would also have a challenge to sync the tracks with the head of the new drive because two drives wouldn't have the same track sync. This meant that you would need a low level format where you only ran the part where track sync was set up. In modern hdds such is done in init when a disk boots up and will track, sector- sync continuously and while reading a sector. If data read from the disk falls under a certain threshold the head will be moved until it is on top of the track again, and this without interrupt the read. Write doesn't have direction error control.
Comments
Post a Comment