|
Table of Contents
1.0.0: Introduction
2.0.0: The Vaio GR300 Series
2.1.0: System Specifications
2.2.0: Hardware Map
2.3.0: Comments
2.4.0: Caveats
2.5.0: Windows Topics
3.0.0: Linux Installation Basics
3.1.0: Allocating Time
3.2.0: Distribution Choices
3.2.1 - What is a distribution?
3.2.2 - Picking a distribution
3.2.3 - SuSE Linux
3.2.4 - RedHat Linux
3.2.5 - Mandrake Linux
3.2.6 - For experts & beyond
3.3.0: Installation Methods
3.3.1 - Full defenestration
3.3.2 - Semi-defenestration
3.3.3 - Proxied defenestration
3.3.4 - Defenestration free
3.4.0: Partition Basics
3.4.1 - Windows Partitions
3.4.2 - Linux Partitions
3.4.3 - PartitionMagic
3.5.0: Performance Basics
3.5.1 - Optimize the BIOS
3.5.2 - Disks and Partitions
3.5.3 - Optimize Partitions
3.5.4 - Filesystem selectiom
3.5.5 - Partition sizes
4.0.0: Installing SuSE Linux
4.1.0: Booting the Installer
4.2.0: Adding Kernel Modules
4.3.0: The YaST2 Installer
4.4.0: Language Selection
4.5.0: Installation Settings
4.6.0: Perform Installation
4.7.0: The First Boot
4.8.0: Handing over the Reins
4.9.0: Final Musings
5.0.0: Appendices
5.1.0: Screen shots
5.2.0: Resources
|
3.5.0 -
Performance Basics
3.5.1 - Optimize the BIOS
The very cornerstone of any computer is its BIOS (Basic Input/Output
System). In order to get the best performance out of your computer, you
should always make sure that its BIOS is set up correctly. BIOS also
controls such things as device access, boot order, and hardware level
security. Although this is changing, and laptop computers feature less
of it, specific hardware components are also configured in BIOS.
3.5.2 - Disks and Partitions
Next to an optimized BIOS configuration, one the most important factor
impacting system performance centers on the data access and storage
subsystem. Although it includes all storage media, the primary emphasis
is on Hard Disk(s) since all significant data transactions executed by
the computer rely on them. Although desktop systems can hold several
Hard Disks, laptops are usually limited to one such permanently
installed device. In either case, special attention needs to be paid to
partitioning, filesystem format and space allocation, and the
limitations of each.
Cylinders, Sectors and Heads - This refers to the method
by which specific locations on the physical disk are addressed. In
essence, the useable surface of the disk (platter) is logically divided
into a series of concentric circles (cylinders), each of which is
assigned a numerical value. Similarly, the platter is also bisected
radially. The areas defined by where these logical dividers intersect
(sectors) are assigned a numerical value as well. Combined, the
cylinder and sector values identifies identify specific locations on
the disk surface.
A Hard Disk contains more than one platter, so there is one mechanical
component that performs the task (head) for each, so in order to read
and write data on the disk, the head is moved to the requested location
of the platter in order to perform the transaction. Together, the disk
location and the head are combined into physical addresses.
The total number of Cylinders Sectors and Heads defines the Hard Disk's
geometry. These values are expressed in that mysterious nomenclature
identified simply as C/H/S. When in doubt about a Hard Disk's
particulars, knowing its geometry will answer most, if not all
questions. It is also required in cases where the geometry must be
entered manually in a system's BIOS.
In order to address the vast storage capacities of modern hard disks,
the addressing is no longer limited to a 1:1 ratio. Instead, a process
known as address translation is performed jointly by the Hard Drive and
the computer's BIOS. Translation groups blocks of addresses for the
Operating System to work with, most notably when the system boots.
Unless a Hard Disk's documentation indicates the contrary, the geometry
always refers to the translated value.
Limitations - The associated hardware only permits a
maximum of four Primary partitions on any Hard Disk, which is somewhat
restrictive. However, it is possible to split a Primary partition into
smaller, logical units, which offers additional flexibility to modern
Operating Systems.
There are Operating Systems which have to be installed wholly in a
Primary partition, as well as some which require a Primary partition
in order to boot. Microsoft Windows is one of the latter, but things
are even more specific here, because Windows goes further by
stipulating that it can only start if it is on the first Primary
partition of the disk. Other Operating Systems may be installed in
either primary or logical partitions, but even they are subject to
limitations (albeit less restrictive ones). Linux may be installed and
started from either a Primary or Logical partition, but the boot
partition has to reside, in its entirety, below Cylinder 1024 of the
disk.
3.5.3 - Optimize Partitions
In order to maximize system performance, the greatest benefit comes
from placing data which is accessed most frequently, first. In the case
of Linux, this will be the bootable partition and the swap partition.
If more than one Hard Disk is available, additional performance gains
will be had by creating multiple partitions on all disks in order to
split the total load.
3.5.4 - Filesystem selection
Another significant source for boosting performance is filesystem
selection. Depending on partition size, some are more efficient than
others.
Windows Filesystems - Microsoft offers two filesystems
for Windows, NTFS and the File Allocation Table (FAT) family of
filesystems. Of these, by far the most efficient is NTFS, but it is not
fully compatible with Linux at this time, and is not designed to work
with boot loaders other than Microsoft's own. The FAT family of
filesystems consists of several types, all of which are Linux
compatible. Of these, the most efficient is known as FAT32, and is the
only other option when used with Windows 2000 or Windows XP. If your
system will be dual-booting, the use of FAT32 for the startup position
is fairly mandatory.
Linux Filesystems - Linux has at its disposal a
significant number of filesystem formats. The most tried and true is
called ext2fs (Second Extended Filesystem) and is akin to a traditional
Unix filesystem. It is comparatively slow and gets progressively more
inefficient as partition sizes get larger, and best suited for small
partitions. Although you're not likely to notice this in day to day
use, the impact is significant during periodic integrity checks, or if
you need to recover data.
Linux is also able to make use of highly advanced Journaling
Filesystems (JFS). These filesystems are a lot more tolerant when it
comes to recovery functions, and become progressively more efficient as
partition sizes increase. Nothing beats a journaling filesystem in
terms of reliability, because of the way it handles integrity checks
and data recovery. There are several such filesystems to choose from,
the two most popular being ext3fs and ReiserFS:
- ext3fs: Developed by RedHat, this
JFS is backwards compatible with ext2fs because it is an overlay to
its antecedent. Because it is an overlay, it is not as efficient as
other journaling filesystems. It really shines in those cases where
it can supercede those aspects which burden ext2fs with
inefficiency, and is ideal if you wish to add JFS support to a pure
ext2fs system since conversion is nondestructive. Note: If you
choose to install RedHat Linux on your machine, this is the only
filesystem type you can choose from, other than ext2fs.
- ReiserFS: The Reiser filesystem
is the result of an independent project created to bring JFS to
Linux. It is highly advanced, efficient, one of the fastest
available, and unlike ext3fs, it is a native filesystem (i.e. not
an overlay). Most distributions will default to this filesystem
format for partitions larger than 30 megabytes. Unless you have a
predilection for a different JFS, it is a safe choice to make.
In addition to the user-interactive filesystems, Linux utilizes one
other, specialized filesystem. This filesystem is used exclusively on
the disk partition(s) which will be used as paging space for the
running system. This partition is required by Linux, so it must be
created if you partition manually, and retained if your distribution's
installer creates one for you.
3.5.5 - Partition sizes
As long as there is sufficient disk space and the Operating System
restrictions can be met, partition sizes are for the most part,
irrelevant. The only partitions for which attention to size is
mandatory for, are the paging space (swap) partition and the startup
(boot) partition.
There is only one constraint involving swap space. Its total size
should be no less than equal to the amount of memory installed in the
computer, and the ideal size is one and a half times that of installed
memory. This space is incorporated into the memory pool of the running
system, increasing the available resources. For instance, a computer
with one gigabyte of RAM installed should have no less than one
gigabyte of disk set aside as a swap partition, with the ideal being
1.5 gigabytes. This results in a pooled total of two to two and a half
gigabytes. The disk allocation can be in the form of one or several
partitions.
The Linux boot partition has to end below the Cylinder 1024 boundary.
If it doesn't, there's a good chance that Linux will not start at all,
or fail during the startup process. The safest way to insure the
requisite is met, is to create a small partition for Linux to start
from. Only the kernel and a handful of support files reside on this
partition.
Although the implied minimum size is one cylinder or an equivalent that
yields at least 24 megabytes, the ideal size should not be less than 50
megabytes. This will allow for the installation of multiple kernels,
which may actually be a necessity if customization is required for the
system to run properly, or for testing newer kernels and still have the
means to roll back to an older one. Anything over 100 megabytes for
this partition can be considered excessive.
 Top of Page
|
|