Chapter 6

Minimum hardware requirements for some distributions include 1 GB of RAM.

Memory

​Replace this with your body text. Have a nice day. Happy teaching!

​​CPU

The minimal amount of drive space recommended is 10 GB. But...

Storage Space

While most servers will have a NIC, multiple NICs for a server are recommended.

NIC

Most racks come with one or more Storage Area Network (SAN) devices. This provides large storage for servers and the network.

​SAN

A Network Attached Storage is similar to a SAN, however, the OS is integrated into the appliance.

​NAS

An Uninterruptible Power Supply (UPS) ensures devices don't loose power in the event of a power failure.

​

​Uninterruptible Power Supply

With a Standby UPS, power is fed to the server directly from the power supply. When power failes, it switches to the battery backup.

​​Standby

With an Online UPS, power is fed to the battery directly from the power supply and the server always draws it's power from the battery. No switching is required.

Online

Virtualization is the process of running more than one OS at the same time on a single system.

​

​Virtualization

There are two types of hypervisors, Type 1 and Type 2.

​

Hypervisors

A Type 1 Hypervisor interacts with the hardware directly and contains a small OS to manage the hypervisor configuration and the VMs. This type of Hypervisor is called a "Bare Metal Hypervisor."

Type 1

A Type 2 Hypervisor requires a host OS in order to provide services to the VMs. This type of hypervisor is called a "Hosted Hypervisor."

​​Type 2

KVM

This is an open source virtualization technology built into Linux. Specifically, KVM lets you turn Linux into a hypervixor that allows a host machine to run multiple, isolated virtual environments called guests or virtual machines (VMs). KVM turns Linux into a Type 1 Hypervisor.

QEMU

QEMU is a generic and open source machine emulator and virtualizer. It can be used in several different ways. The most common is for System Emulation or work with a hypervisor such as KVM. The framework allows the guest to run directly on the host CPU.

Legacy SCSI was developed in the 1980's and relied on ribbon cable to transmit information between the disk and controller in parallel. Each device used a unique Logical Unit Numbers (LUN)

SCSI

SAS is the next generation SCSI and allows for data transfer speeds greater than it's predecessor. Once connected to the controller, all configurations are performed automatically.

​

SAS

RAID allows for a variety of configurations for fault tolerance, speed, and capacity.

​There are 7 RAID levels but only 4 are widely used.

RAID

RAID 0, also called Striping, is spanning across two or more drives so the system sees one large volume.

​​RAID 0

RAID 1, mirroring, requires an even number of disks so as data is written to one, it is copied to the other. This is for Fault tollerance

RAID 1

RAID 5 is striping with parity. The parity bit is used to rebuild the array in the event 1 drive failes. This requires at least 3 drives.

RAID 5

RAID 6 is the same as RAID 5, but adds 6 parity disks and allows for 2 drives to fail.

RAID 6

RAID 10 combines RAID 1 (mirroring) with RAID 0 (stripping) to maximize storage and provide fault tolerance to the array.

RAID 10

Two ways to handle RAID configurations, Software RAIDs and Hardware RAIDs. Hardware RAID is more common.

RAID Configurations

Using the MDADM Package

Linux can configure a RAID during installation or after installation using the mdadm package. The package used the multiple disk (md) device file to manage the disks associated with the RAID level chosen. The mdadm package hides the member drives from Linux so the OS only sees the md device.

ZFS

ZFS is a local file system and logical volume manager created by Sun Microsystems, Inc. to direct and control the placement, storage and retrieval of data in enterprise-class computing systems. It has scalablility, data integrity, and many built-in functions that make it well suited for enterprise systems.

• Replication

• Deduplication

• Compression

• Snapshots

• Clones

• Data protection

ZFS Provides

ZFS

ZFS uses "pools" of disks (zpool) to that can be managed and volumes (ZFS managed filesystems) can be created from these pools. Two primary commands to use are:

• zfs - Manages specific features of the ZFS filesystem

• zpool - Creates and monitors disks added to a ZFS filesystem.

BTRFS is a modern copy on write (COW) filesystem with advanced features that also focus on fault tolerance, repair and easy administration.

B-Tree File System

BTRFS

BTRFS supports:

• Snapshots which do not make a full copy of the files

• Built-in volume management, support for software-based RAID 0, RAID 1, RAID 10 and others

• Self-healing - checksums for data and metadata, automatic detection of silent data corruptions

Which Distribution to Use?

In the Windows world, in order to provide access to large numbers of users, the server needs to run the Windows Server OS and have the correct number of Client Access Licenses (CAL) installed.

In Linux, any distribution can become a Linux server simply by installing necessary packages to perform the tasks needed.

Any distribution can be a server

Any Linux distribution can be configured to be a server. Since all Linux distributions support the same software packages as their server siblings, any distribution of Linux can be configured to be a server as long as the hardware will support it.

During installation, workstation and server distributions install a pre-defined set of packages.

Log files are an excellent location to locate errors after installation or any time a problem occurs.

Log Files

If a device is not working properly, there are many tools that can be used to determine the problem.

​

Device Problems

Another method to check if device drivers are loading is to check that the correct modules are loading. However, some device drivers may have been compiled into the kernel.

Modules

journalctl command

The journalctl command is another excellent tool to use to determine errors, especially during boot. The -b option can be added to view all of the boot messages listed in the journal.

journalctl -b --since=today

There may be times when you need to rescue a system. This can be due to boot problems, kernel problems, or configuration problems.​

System Rescue

Also having a rescue disk available will help to repair a broken system. Many of the distributions that have a Live OS option will allow you to repair a broken distribution.

​

​Rescue Disk