1

Servers
Chapter 4
 What is server? 2

• Servers are specialized computers designed to

provide services or resources to multiple
clients.
• Unlike workstations that serve individual users,
servers cater to numerous clients
simultaneously.

• They are the backbone of modern IT

infrastructure, supporting various applications,
websites, databases, and network services.
 Features of Server Hardware 3

Server hardware usually costs more but has additional features that
justify the cost. Some of the features are

o Extensibility:
Servers offer room for growth, accommodating more hard drives,
cards, and CPUs for enhanced performance and capabilities.
 Features of Server Hardware 4

o Enhanced CPU Performance:

Multiple CPUs, advanced features, and dynamic resource allocation ensure optimal
server processing power.

o High-Performance I/O:
Servers handle extensive input/output operations with faster disk drives, internal
buses, and high-speed network interfaces.

o Upgrade Options:
Server CPUs can be upgraded without major hardware changes, ensuring scalability.
 Features of Server Hardware 5
o Rack-Mountable:

Servers are designed for efficient rack installation,

space utilization, and optimized cooling.

o No Side-Access Needs:

Maintenance can be performed without removing

servers from racks, saving time and effort.
 Features of Server Hardware 6
o High-Availability Options:

Redundant power supplies, RAID, multiple network connections, and hot-

swappable components enhance reliability.

o Maintenance Contracts and Management Options:

Vendors offer service contracts for timely repairs, and remote management tools for
diagnostics and problem-solving.
 Choosing Reliable Vendors for Your Servers 7

o Selecting reliable vendors is paramount when sourcing server hardware.

o Some vendors prioritize quality, while others may compromise by using
consumer-grade components or those meeting MIL-SPEC requirements.
o Experienced vendors bring years of server design expertise and offer
additional valuable features that come from market experience.
o Inexperienced vendors may lack essential maintenance services, often
only providing exchanges for defective units.
 Choosing Reliable Vendors for Your Servers 8
o Seek recommendations from fellow System Administrators (SAs) and leverage
resources like the System Administrators’ Guild (SAGE) and the League of
Professional System Administrators (LOPSA) for vendor insights.
o Consider your environment type: homogeneous (same vendor/product line) or
heterogeneous (multiple vendors/product lines).
o Homogeneous environments simplify training and maintenance but limit vendor
choice.
o Heterogeneous environments offer flexibility and competition among vendors,
potentially leading to better service (Chapter 5 discusses this further).
 Understanding Server Costs 9

• Servers come in different types:

• home, business, and server lines.
• Home servers offer a lower initial price but may require costly upgrades.
• Business servers focus on long-term value and stability, reducing
component changes.
• Servers prioritize cost-effectiveness and uptime, justifying their price.
 Understanding Server Costs 10

• Features like easy maintenance and redundancy add to server costs but
reduce downtime.
• While servers seem pricey compared to high-performance PCs, they
provide reliability and efficiency.
• Overpowered servers can delay future upgrades, so consider your needs
carefully.
 Maintenance Contracts and Spare Parts 11

1.Maintenance Contracts: Consider various maintenance contract options when buying

servers, based on criticality and response times.

2.Spare Parts Kits: Evaluate the need for spare parts kits to expedite repairs and ensure
they contain one part for each component in use.

3.Cross-Shipping Parts: Opt for vendors that offer cross-shipping of replacement parts
for faster repairs.

4.Regular Contract Review: Regularly review service contracts to add new servers and
remove retired ones, ensuring accurate coverage and cost management.
 Maintaining Data Integrity 12

• Servers store critical data and unique configurations, necessitating robust

protection measures.
• Workstation clients often share identical configurations and rely on server-
based data storage.
• Despite best practices, users may store data locally, install software, or
retain OS configurations.
• UNIX systems, with proper configuration, offer better data protection, but
some files may still require backup.
• Effective backups are essential for safeguarding server data integrity.
 Placing Servers in the Data Center 13

• Servers belong in environments with proper power, fire protection,

networking, cooling, and physical security.
• Allocate server space when purchasing to avoid double-booking.
• Mount servers in the rack immediately after hardware assembly to prevent
premature heavy usage.
• Small offices should have designated spaces with basic infrastructure like
UPS and cooling.
• Even telecom closets with secure cooling are preferable to servers under
desks for data protection.
 Client Server OS Configuration 14

• Versatile Server OS: Servers can run different, identical, or slightly

modified OS configurations depending on their intended use.

• Protocol Compatibility: Servers, like web servers, only need to agree on

protocols with clients, allowing for diverse OS choices.

• Specialized Appliances: Some servers, such as network appliances, utilize

minimal OS configurations tailored to specific functions.
 Client Server OS Configuration 15

• Uniformity in Environments: In scenarios like UNIX environments with

multiple desktops and CPU servers, uniform OS loads may be preferred.

• Perspective Matters: Server OS appropriateness varies by perspective;

some OS installations may load all software packages, while others opt for
minimal configurations with additional software added as needed.
 Remote Console Access 16

• Servers need remote console access for maintenance.

• KVM (Keyboard Video Mouse switch)switches and IP-KVMs save space and
allow remote control.
• Serial console concentrators are also valuable for remote management.
• Remote access enhances productivity and safety while reducing machine
room congestion.
• Security measures like SSH and authentication systems should protect
remote access.
 Mirror Boot Disks 17

• Boot disks containing the OS should be mirrored (RAID 1).

• Mirroring provides redundancy in case of disk failure.
• Although writes may be somewhat slower, mirroring enhances read
performance.
• Disk mirroring allows for controlled disk replacements without
downtime.
• Mirroring safeguards against hardware failures but doesn't protect
against software or human errors. Backups are still essential.
 Enhancing Reliability and Serviceability 18

• Optimizing Server Management

In today's fast-paced digital landscape, ensuring your servers operate at peak

performance and can be quickly maintained or repaired is crucial.
 Enhancing Reliability and Serviceability 19

• Server Appliances
• Specialized devices are designed for specific tasks, such as
• file servers, web servers, email servers, and DNS servers.

• General-purpose servers, appliances excel at their designated roles,

simplifying deployment and management.

• They're particularly valuable for organizations lacking in-house

expertise in server architecture.
 Enhancing Reliability and Serviceability 20

• Redundant Power Supplies

After hard drives, power supplies are among the most

failure-prone components in a system.
• To mitigate this risk, servers should feature redundant
power supplies, meaning the system remains
operational even if one power supply fails (n + 1
redundancy).
• Ensuring each power supply has a separate power cord
and, for high-availability systems, draws power from
different sources (e.g., separate UPSs) enhances
reliability.
 Enhancing Reliability and Serviceability 21

• Hot-Swap Components

Hot-swappable components enable the removal

and replacement of server parts while the system
remains operational.
• This feature is invaluable for reducing downtime
in n + 1 redundancy configurations, where a failed
component must be replaced promptly to
maintain system reliability.
• Examples include hot-swappable hard drives and
RAID systems with hot spare disks.
 Enhancing Reliability and Serviceability 22

• Separate Admin Networks

Additional network interfaces in servers enable the creation of separate

administrative networks.

• These segregated networks are beneficial for tasks like backups and
monitoring, as they prevent these processes from affecting the main network's
performance.

• Furthermore, they provide a means for system administrators to access servers

during network outages, enhancing reliability and serviceability.
 Enhancing Reliability and Serviceability 23

• Many Inexpensive Servers

Instead of investing in high-end, expensive servers, consider using numerous

low-cost servers with replication and load balancing.

• When effectively managed, this strategy can be more cost-effective and

scalable, offering redundancy and performance improvements.

• This approach has revolutionized large web services like those operated by
tech giants such as Google and Yahoo!.
 Enhancing Reliability and Serviceability 24

• Disposable Servers

In certain scenarios, it might be more economical to leave non-functional

servers in place rather than repairing them.

• Removing these servers could inadvertently lead to network disruptions due to

cable disconnections.

• We examine a real-world case study highlighting the challenges and solutions

associated with managing disposable servers in a large-scale environment.
 Enhancing Reliability and Serviceability 25

• Blade Server Management

Blade servers are compact, high-density solutions with multiple slots for CPU
and memory cards. They simplify server management and maintenance,
particularly when used in clusters