7. Managing Partitions
The Primary partition is the part of the disk that can be used to start the computer. A hard disk
can have 4 primary partitions. The NT system partition must be a primary partition. There can only be
one Extended partition on a hard disk. The NT boot partition can be in the primary partition or
a logical drive in an extended partition.
A Volume Set is a collection of free-space areas combined into a logical drive, it does not improve
performance. These free-space areas can come from SCSI, IDE and ESDI drives and consist of 2 to 32
different areas. Boot and system partitions cannot reside in a volume set and there is no fault tolerance.
Only an NTFS volume set can be extended.
A Stripe Set is very similar to a volume set other than they require at last two physical disks.
The data is written evenly in 64Kb chunks across all the disks. If the controller can access
numerous hard disks simultaneously then a stripe set can cause a system to speed up. Windows 95
does not recognise stripe sets.
Disk Administrator, in the Administrative Tools Tools folder,
is used to manage hard disks. This program and Ftdisk.sys, which
is the Fault Tolerance Driver create a 32 bit signature for each disk.
Partitioning and formatting takes place from Disk Administrator.
The Active partition is the primary partition containing the system boot files
and cannot be part of a volume set or a stripe set. Only
one partition can be active at a time and this is marked with an asterisk in Disk Administrator.
NT assigns partition numbers to all primary partitions before assigning numbers to any logical drives
within an extended partition. Because Boot.ini is used to find the boot partition, the
boot.ini file must be updated to point to the boot partition if it was in
an extended partition that was renumbered.
NT automatically assigns drive letters starting with the primary partition and the active partition.
These drive letters can be reassigned manually via disk administrator.
Rdisk.exe is used in conjunction with the Emergency Repair Disk to repair problems
with disk configuration information.
On NT Servers you can implement RAID (Redundant Array of Inexpensive Disks) 0, 1 and 5. RAID 0
is a stripe set and has no tolerance. RAID 1 is disk mirroring using the fault tolerance driver
Ftdisk.exe to simultaneously write data to two disks with one controller. Disk Duplexing uses
a controller for each disk and is quicker. RAID 1 can mirror the boot or system partitions.
RAID 5 is Stripe Sets with with Parity. Data integrity is maintained by adding a parity stripe to each
disk partition in the volume. If a disk failed the parity is used to reconstruct the data
on the other disks. Data writes are slower as a result.
RAID 5 is configured from the Fault Tolerance menu in Disk Administrator. If a disk fails
and has been replaced you then need to choose Regenerate from the fault tolerance menu.
Event Viewer is used to show which partition has failed. If a mirror set has failed then break the
mirror set in the Fault Tolerance menu before adding the new disk.
Follow these steps to create a fault tolerance boot disk:
- Fomat a floppy under NT, this automatically puts system tracks on the floppy.
- Copy Ntldr, NTdetect.com, Ntbootdd.sys (for SCSI systems that do not use BIOS), Boot.ini on to the floppy
- Modify the Boot.ini file so that it points to the mirrored copy of the boot partition.
- Test that it works by booting the computer using the floppy.
The Advanced RISC Computing (ARC) naming convention used in Boot.ini is as follows:
- scsi(v) - is used when the SCSI adapter has the BIOS NOT enabled. The numbering (v) starts at 0.
- or multi(w) - is used for SCSI (with BIOS enabled), IDE or ESDI adapters. The numbering (w) starts at 0.
- disk(x) - SCSI bus number, for multi, x is always 0.
- rdisk(y) - number of the disk (ignored for SCSI), y starts at 0.
- partition(z) - number of the partition, z starts at 1.