An introduction to operating systems

1 Introduction
There is no single definition of operating system. Operating systems exist because they are a
reasonable way to solve the problems created by a computer system. The hardware itself is not easy
to use; therefore it is necessary to help both the user and the programmer by abstracting the
hardware complexity. The way to do this is by placing a layer of software above the hardware in
order to present the user of the system and the applications a virtual machine interface that
facilitates the understanding and use of the system. This software layer is called the operating
system. The operating system includes a set of functions to control the hardware that are common
to most applications, e.g., functions controlling the devices and interrupt service routines, hiding to
the programmer the hardware details and offering an interface comfortable to use the system.

From another point of view, the operating system Metaphor of the driver and the mechanic
must ensure a proper and efficient system
In any system it is important to distinguish between
functioning. A computer system now consists of a interface and implementation. The user of a system
large number of components that need to be must know its interface, but how it is implemented is
managed. Throughout the history of computers a matter of the designer or the maintenance staff.
The user of a car only needs to know the interface so
there has been a significant development affecting that the vehicle is helpful. So, he must learn to
the various components of the system. This manage the steering wheel, turn signals, lights,
accelerator and brake. To make things easier,
evolution has been achieved both in the manufacturers tend to standardize the interface: the
technological aspect (from valves and relays to accelerator is a pedal that is always located in the
VLSI circuits) and at the architectural level same place; the direction "right" is always
represented in the controls as a turn in the clockwise
(different techniques to increase processor speed, direction... Since the system is not perfect, the user
memory hierarchies...) and in the field of must perform some "management" tasks: if the car
is not automatic, he must choose the right gear,
programming languages (libraries, languages, when the tank is empty, he must fill it with a given
interfaces...). This evolution was constrained by type of fuel... However, these tasks tend to be
requirements of efficiency and ease of use of increasingly limited. A century ago, the user of the
car used to have a driver-mechanic, since the cars
computers. However, it should be noted that the were very unreliable, and should be started
increased efficiency of each system component manually by operating the starter motor from the
outside. Today, one can be a good driver without
does not ensure an increase in the overall system having knowledge of mechanics, and many drivers
efficiency. Indeed, the tuned management of all ignore, for example, that the car has an electric
resources will be largely responsible for the motor for starting. The mechanics are in charge of
maintenance, knowing perfectly the internal
success or failure. From this perspective, the structure of the car, but they do not have to be good
operating system is responsible for providing an drivers: they might not even know how to drive.
ordered and controlled allocation of different
resources (processor, memory, Input/Output devices...) to each of the programs competing for them.

We can say that the concept of operating system is linked to two different ideas. For a
user/programmer, an operating system is the set of functions that allows him to use the resources of
the machine obviating the characteristics of the hardware. This is the functional vision of the
operating system, which allows one to view the system as a virtual machine. This is the vision that
this course will deepen. For a system designer, however, an operating system is the software that,
installed on the bare machine, allows controlling its resources efficiently. This view corresponds to
the operating system implementation.

Both views largely refer to the same concepts and terms, but their approach and their
objectives are different. In this introductory course to operating systems we will study the

KAT/ATC UPV/EHU
An introduction to operating systems 2

functionalities offered by operating systems in general, as well as the basics of how the operating
system supports them. The techniques and fundamental models of the design of operating systems,
as well as the concepts and tasks of system and network administration, including security
management, are studied in courses of the Computer Engineering specialization.

2 Functional vision of operating systems

Of the two approaches presented out above, this is the less clearly defined and developed in the
literature. Perhaps this is due to the fact that historically it has been the interface programmer who
designed the interface functionality, and he does not feel particularly inclined to discuss the specific
services that the interface must provide. Hence, often services are later added or modified according
to the needs in revisions. In the operating system, in addition, this interface is not unique, in the
sense that, besides the set of system calls (primitives of the operating system) provided to
applications, it can be considered historically it has been the shell as part of the operating
system, and even, by evolution, the graphical user interface (GUI).

In what follows, we will consider the system call interface as the basic interface of the operating
system, which defines the system as a virtual machine. The set of system calls of an operating
system describes the interface between applications
and the system and determines the compatibility Interfaces and interfaces
between machines at the source code level. Perhaps the most frequent professional vice of
computer engineers is to not properly differentiate
the various system interfaces. One can just take a
The end user sees the computer system in terms of look at any application or operating system to
applications. Applications can be built with a realize it. For example, one can find in
programming language and are developed by "Accessories", together with a calculator or a player
and a sound recorder, tools like "disk
application programmers. If we had to develop defragmenter". Surprisingly, in "Control Panel" one
applications taking care at all times of the control of can find an application to read text aloud. It's like if
the hardware they use, application programming a car manufacturer had placed a wrench on the
dashboard, next to the hole for sunglasses, and the
would be a daunting task and probably we could not radio under the hood, near the engine. Not
enjoy sophisticated applications as the ones we have surprisingly, many home users hate computers.
"Computing is very complicated", they say. Well, so
nowadays. In addition, applications also take is mechanics. However, no driver feels unsafe while
advantage of a set of tools and services that driving by not knowing how to use a wrench. In
facilitate even more the work of the programmer, as this sense, one could say that current operating
systems are like cars of a century ago.
editors, compilers, debuggers Here we also
include libraries of functions that are available to
applications (mathematical functions, graphical). Typically, these services are not part of the
operating system. Figure 1 presents a summary of this approach.

End user

Application
Application
programmer
programs
Tools
OS OS
designer
Hardware

Figure 1. Layered structure of a computer system.

KAT/ATC UPV/EHU
An introduction to operating systems 3

3 Functions of an operating system

In general, and regardless of the type of interface, operating systems typically provide a set of
functions that can be summarized as follows:

Program execution. Running a program requires a number of tasks. Instructions and data
must be loaded into main memory, files and I/O devices must be initialized The operating
system performs all these tasks.

Control of I/O devices. Each device requires its own set of instructions and control signals
to operate. The operating system takes care of all these details so that the programmer can
see the access to the devices as simple reads and writes.

Access to files. Historically we have used the concept of a file as the permanent
representation of a set of information with a global name in the system. Files reside in
nonvolatile memory such as disks and flash drives. Besides the nature of the device, the
operating system has to manage the file format and the way of storing.

System access control. For multi-user systems, the operating system has mechanisms to
control the access to system resources based on the rights defined for each user.

Detecting and responding to errors. When a computer system is in operation it may fail.
These errors can be hardware (memory or device access error), or software (arithmetic
overflow, attempt to access a forbidden memory position...). In many of these cases the
system has hardware components to detect these errors and to communicate to the operating
system, which should give a response that eliminates the error condition with the least
possible impact on the applications that are running. The answer may go from the ending of
the program that caused the error, to retrying the operation or simply reporting the error to
the application.

Accounting. It is common for an operating system to provide tools for tracking operations
and accesses, and for collecting data regarding resource usage. This information may be
useful to anticipate the need for future improvements and to adjust the system so as to
improve its performance. It can also be used for billing purposes. Finally, upon a security
issue, this information can be used to discover the attacker.

4 Operating system interfaces

In a system structured in layers, a layer Lk provides an interface to the upper layer Lk+1, represented
by a set of functions which determine how layer Lk is accessed from layer Lk+1. The implementation
of layer Lk is independent of the interface and is said to be transparent to the layer Lk+1, in the sense
that when designing the layer Lk+1 there is no need to worry about how layer Lk is implemented. An
interface must specify precisely the functions offered and how they are used (arguments, return
values).

Generally, an operating system offers three different interfaces:

User interface. When there were no graphics terminals like the ones we have nowadays, the user
had to communicate with the system by typing commands that allowed running programs,
consulting directories To do so, the operating system offered a specific utility, the command
interpreter (shell in Unix terminology), whose interface was presented as a set of commands whose

KAT/ATC UPV/EHU
An introduction to operating systems 4

usage form was (or should be) well specified in a manual (for example the Unix man, Section 1).
Nowadays graphical user interfaces greatly facilitate user interaction by means of intuitive concepts
and objects (icons, pointers, mouse clicks, drag and drop...). If in the case of shells each system
offered its own shell (the user had to learn to use it, usually attending a course), the graphical user
interfaces are common and intuitive enough so that their use is available to everyone.

Administration interface. The administrator of a computer system is the person in charge of

installing the system, maintain it and manage its use. In a system composed of several computers,
this work includes managing user accounts and network resources, with special attention to the care
of user privacy and information security. The system administrator is a professional who knows the
specific tools and functions that the system offers for it and that can only be used by him, as they
require special privileges. Overall, he relies for it on an extension of the shell (for example, in Unix,
specified in Section 8 of man), although the use of these tools does not exclude the use of the
graphical user interface. Instead, a personal system should not require, ideally, management effort
by the user, since he is not supposed to be an expert for it, like the driver of a car is not required to
have mechanical expertise. The reality is that, like a car driver should know how to change a wheel,
a computer user has to solve nowadays some management problems arising from the immaturity
and imperfection of operating systems.

Programming interface. To develop applications on an operating system, the programmer uses,

regardless of the programming language used, a set of functions to access operating system
services, the system call interface. These functions do not differ in appearance from other library
functions provided by the language. However, calls to the operating system are specific to that
system and therefore probably incompatible with those of another operating system, since they refer
to objects and concepts specific to that system.
APIs
Actually, it is common that the programmer does
Nowadays programmers use to talk about API
not directly use operating system calls, but specific (Application Programming Interface) to refer to the
library functions of the language for that purpose. set of functions available in a platform for
For example, if the C programming language is application development. An API can be the set of
system calls extended with other library functions,
used, the programmer uses the printf function to though system calls themselves are usually hidden
output data, regardless of the operating system he by library functions that facilitate programming.
There may also be specific APIs tailored to specific
is using. However, printf is a function applications. Ultimately, an API depends on the
implemented in terms of calls to the operating programming language and operating system for
system (in the case of Unix, the write system call), which this API is implemented.
so that the code generated is specific to that In the Java world, since it is an interpreted
system. This, in general, is not taken into account language, the APIs are independent of the operating
system: it is the virtual machine (JVM) which
by the application programmer, but it is by the interprets the library functions for the underlying
library developer, a systems programmer, who is operating system.
the user of the operating system call interface and
will therefore rely on the corresponding specification (in Unix, Section 2 of man).

5 Evolution of operating systems

From the perspective offered by the already relatively long history of operating systems, and
considering its application fields, now we can talk about different models of computation, which
determine the functionality of an operating system, and sometimes its structure:

Batch systems. The earliest operating systems (1950s) were called monitors. The users gave their
program with the input data in a stack of punch cards (a lot) to the computer operator who
sequentially ordered lots and placed in a card reader. Each batch included control cards with orders

KAT/ATC UPV/EHU
An introduction to operating systems 5

for the monitor. The last card was a return order to the monitor that allowed it to start automatically
loading the next program.

Multiprogrammed systems. The price of a CPU at that time was exorbitantly high, so it was
intended to work 100% of the time, which is unattainable with batch systems, since the processor,
when executing an I/O instruction, should wait for the device, very slow compared to the processor
speed, to complete the operation. This led to the engineers of the time to devise strategies for a
more efficient use of the CPU. By loading multiple programs in memory, when a program needed
to wait for an I/O the processor could execute another program. This technique, known as
multiprogramming or multitasking, was developed in the mid 1960s and is the basis of modern
operating systems.

Time-sharing systems. At that time new applications appeared requiring an operating mode in
which the user, sitting at a terminal, interacted directly with the computer. This operating mode,
interactive, is essential, for example, in the processing of transactions or queries. The interactive
processing requires, of course, multiprogramming, but must also provide a response time (time
elapsed from the ordering of a transaction until the answer is obtained) reasonably short. That is, the
user that interacts from a terminal can not be waiting for long because some program, aimed at
calculation, does not leave the CPU for not executing any I/O for a while. For this reason, in time-
sharing systems, introduced in the second half of the 1960s, the operating system runs the
programs in short bursts of computation time (quantum), in an interleaved way. Thus, if there are n
programs loaded in memory, each program will have in the worst case (when no program required
I/O) 1/n of the processor time. Given a quantum small enough and a not too big n, the user does not
observe a significantly long response time for his program and has the feeling of being using a
dedicated processor with a speed 1/n of the actual processor. This idea is known as shared
processor, and reflects the ideal behavior of a time-sharing system, minimizing the response time.

Teleprocessing systems. In the first time-sharing systems terminals were connected to the
processor by means of specific wiring that was installed in the building. When large companies and
institutions (e.g., banks) began buying computers, they found the need to transmit information
between their branches and the computer at the headquarters. Fortunately, there already existed the
telephone wiring, which was used to transmit
A question of price
digital information using a modulator-
demodulator (modem) at each end, connected to It is necessary to look at the evolution of the cost
factor regarding technology to understand the path
the conventional telephone line. Unlike the followed by the system management models.
transmission using special wiring, telephone Before the development of integrated circuit
technology, a computer costing millions of dollars,
communication is very prone to errors, so it was was composed of tens or hundreds of thousands of
necessary to develop more sophisticated individual electronic components (transistors and,
communication protocols. These protocols were, previously, valves), weighed several tons and
occupied a large and heated room. However, their
initially, proprietary (owned by the computer performance in terms of processing power and
manufacturer, which was also the one who storage were comparable to those of the chip in a
smart card of today. It can be understood then that
supplied the terminals, modems and software). in the 1960s, engineers started the development of
operating systems with multiprogramming and
Personal systems. Cheaper hardware and the virtual memory, able to take full advantage of these
machines (IBM/360s basic configurations, the most
advent of the microprocessor in the late 1970s popular mainframe of that era, came with 8 Kbytes
made it possible to provide a dedicated system for of memory and executed a few thousand
a single user at a reduced cost, a key feature of a instructions per second, yet the CPU was very fast
compared to the punch card reader). Today,
personal system. The operating system for operating systems still include virtual memory, but
personal computers is, at first, single-user (no most personal computers do not need to use it.
protection mechanisms) and single-tasking, that
is, not very different from the primitive monitor-based systems except for the fact that it is used

KAT/ATC UPV/EHU
An introduction to operating systems 6

interactively through a terminal. Today the available hardware allows multitasking personal
systems (Mac OS, Windows, Linux) supporting sophisticated graphical user interfaces.

Networked systems. With the advent of the personal computer the terminals of teleprocessing
systems are replaced by PCs that can take certain computing tasks, downloading the central time-
sharing system. In particular, PCs can execute any communication protocol. With the adoption of
standard protocols (e.g., TCP/IP), personal computers can communicate with each other: there is no
one central computer, but a set of computers that are connected together. If a computer in the
network provides access to a particular resource, then it is the server of that resource. The
remaining computers, clients, access the remote resource using a client-server protocol. Managing
access to networks has complicated the operating system and has led to the emergence of services
that are deployed on it (known as middleware), resulting in distributed systems that are deployed
today in the field of Internet and have generated concepts and schemes very sophisticated, such as
Web services, peer-to-peer and cloud computing. Although this course is restricted to the study of
centralized systems, we must not forget that the reality is more complex.

Mobile systems. The evolution of hardware does

Groundhog Day (1993)
not end with personal computers. These are
The long history of operating systems has followed
becoming smaller, which, together with the use of a a cyclical path. It is surprising to learn that
battery and a wireless network, provides autonomy sophisticated concepts and complex techniques to
and makes them mobile systems. In principle, this implement such as multiprogramming and virtual
memory have almost half a century and were part
change does not significantly affect the operating of the first time-sharing systems. When, fifteen
system. However, with the new century and by years later, personal computers appeared, the first
operating systems developed for them dispensed
means of the evolution of mobile telephony new with these mechanisms because their limited
devices with increasing computing capabilities have hardware could not support them. In fact, apart
appeared. These devices, now called smartphones, from the interactive operation mode, they were not
very different from the primitive monitors.
are capable of supporting smaller versions of However, as the hardware of personal computers
operating systems designed for personal computers gained in performance, their operating systems
(Mac OS, Windows, Linux), although there are also were integrating these techniques. So, while at the
time they distinguished between mainframes,
specific operating systems (as Symbian, or Google workstations and personal computers, any
Android) with great performance, including new computer today is capable of supporting a complex
operating system. More recently, miniaturization
forms of interaction (touch screens, cameras, has led to the emergence of small devices (mobile
positioning information...) and new applications phones, smartphones, are the most notable
(such as navigation). This field is undoubtedly the example) with increasing computing and storage
capacity. Again, history is repeating itself: if the
hottest area for the development of current and first operating systems for mobile phones were
future technology of operating systems and extends extremely simple, there are already smaller
versions of general purpose operating systems
to very different types of devices (e.g., cameras, aimed at mobile phones, which are integrating
smartcards, or control devices embedded in features like multitasking.
appliances or cars...), which are capable to network
and interact spontaneously with each other even without human intervention.

6 A classification of operating systems

When classifying current operating systems one can take into account different criteria, derived
from the concepts introduced above. One possible classification is based on the following criteria,
which can be combined: (1) if the system can run at the same time one or more than one program,
(2) if it supports the connection from a single terminal or from more than one, and (3) whether it
supports a single user or can manage more than one user.

KAT/ATC UPV/EHU
An introduction to operating systems 7

(1) Monoprogrammed/multiprogrammed. They are also known as single-tasking/multitasking,

terms that we will consider synonymous. In the primitive operating systems, both monitors as the
first systems for personal computers, for example MS-DOS, the execution of a program had to
finish for the start of the next program. These systems are called monoprogrammed (single-
tasking). From 1965 there appeared the first multiprogrammed systems (OS/360, Multics). Today,
virtually all operating systems are multiprogrammed (multitasking). In multiprogrammed systems,
several programs run concurrently, i.e., interleaving their executions over time, which are
perceived as simultaneous. They use the concept of process (or task) to designate a running
program. As stated above, multiprogramming was motivated by the need to optimize processor
usage, and therefore running processes in a multiprogrammed system usually represent independent
applications. Later multiprogramming has been used to express concurrency in the same
application, where a set of tasks cooperate in a coordinated manner. For example, in a word
processor we can find a task in charge of reading and processing keyboard input, another task in
charge of checking the spelling, a third task responsible for periodically saving changes A
particular class of multiprogrammed operating systems is the multithreaded systems, which allow
expressing the concurrency in an application more efficiently. The difference between a process and
a thread (also called subprocess) is, for our purposes, very small, and we will not address it at this
time. Thus, multiprogramming means multiplexing the processor among processes, as explained
above. Obviously, a multiprocessor system (a computer with multiple processors) enhances further
the multiprogramming by allowing the concurrent execution of programs to be also parallel. This
is known as multiprocessing, and operating systems that control these systems are called
multiprocessor operating systems. Although there are significant differences in the
implementation of a multiprocessor operating system with respect to a single-processor operating
system, with respect to the functional vision of applications and users they hardly transcend.

(2) Single-terminal/multiterminal. An operating system ready to be connected simultaneously

from different terminals is said to be multiterminal, otherwise it is said to be single-terminal. Time-
sharing operating systems, such as Unix, are multiterminal. An operating system designed for
personal computers MS-DOS, Windows 95/98 is, naturally, single-terminal. It is noteworthy
the case of Linux, a Unix system for personal computers, which maintains the multiterminal Unix
philosophy by means of a set of virtual terminals. Mac OS X, also derived from Unix, is another
multiterminal example. It is clear that a multiterminal system must be somehow multiprogrammed:
as we shall see, it is common that each terminal (real or virtual) has an associated process that
manages the connection.

(3) Single-user/multiuser. A multiuser system is able to provide user authentication and includes
policies for managing user accounts and access protection, providing privacy and integrity to users.
In the primitive monitor-based operating systems, shared by several users, this function was carried
out manually by the system operator. The first operating systems for personal computers, such as
MS-DOS, were single-user. The general purpose operating systems of today are multiuser. Note
that some personal systems, such as mobile phones, include some verification mechanism (usually a
password), but lack of policies to protect accesses to system resources and user management; they
simply authenticate the user, but are in all aspects single-user.

7 The operating system market

From a closer perspective to the business world, we must refer to two groups of operating systems.
First, those operating systems that have been designed by a manufacturer for a specific architecture
in order to protect their products (both software and hardware) for potential competitors, which are
called proprietary operating systems. The manufacturer designs the operating system specifically

KAT/ATC UPV/EHU
An introduction to operating systems 8

for the architecture, and provides the necessary updates. Even sometimes the specification of its
system call interface is not made public or is constantly changing, making difficult the development
of applications by other manufacturers. This creates a closed world that encompasses the
architecture, the proprietary operating system and the applications, enabling the control by the
manufacturer of the market for their products and establishing big dependencies for customers.
Some examples of proprietary operating systems, largely deployed, are (or have been) IBM
systems, Digital VAX VMS, Apple Mac systems, and Windows systems of Microsoft for the PC
platform.1

With the advent of Unix (circa 1970) a new philosophy arises: since it is written almost entirely in a
high level programming language (C), the operating system is portable to other architectures and
therefore so are the applications at the source code level. Furthermore, in the case of Unix, its
source code was freely distributed. This had contradictory effects: on the one hand it contributed to
the wide dissemination of the system; on the other hand, each manufacturer introduced their own
modifications not only in the source code but also in the system call interface, so that you have to
refer to different Unix systems, not fully compatible with each other (System V, BSD, AIX, Ultrix,
Solaris, Linux...). We can say that the family tree of Unix is really complex.

The ideal consisting of a world of open systems, with public specifications, accepted and
standardized, allowing full portability of applications (and users2), is a goal rarely achieved. In this
regard, there have been efforts to define standard specifications. For example, the POSIX
specification is a reference in the Unix world. A developer that follows in the system calls of its
program the POSIX specification knows that he can compile and run it on any Unix system that
follows the POSIX standard.

In this sense, it would be useful that operating systems were designed with the ability to support
different system call interfaces. This was the philosophy of microkernels in the 1980s, which
implemented the system call interfaces as services outside the operating system itself. However, the
development of microkernel-based operating systems has had a limited commercial impact. The
best known is the Mach 3.0 microkernel, on which the Mac OS X operating system from Apple
relies. However, the most common approach today is to support applications of heterogeneous
systems through emulation (virtualization) of other operating systems on a host operating system.
There are numerous virtualization programs, e.g., VMware, Virtual PC, or Win4Lin.

It should be noted a phenomenon that revolutionized the market of software in general and
operating systems in particular: the spontaneous emergence of a community of programmers who
develop free software3. Internet is the necessary way for sharing and exchanging code and ideas
rapidly in the community. As a result, and this has been amply demonstrated, the software adapts
very dynamically to particular problems, the development of new products is very fast, and errors
are corrected and versions refined with great agility. Organizations like GNU4 grant a license to
copy, modify and redistribute free software with the condition that the new distribution includes the
source code.5 Linux is today a settled example of this philosophy.

1
Still and all, there are important differences between proprietary systems. For example, Microsoft had the success in
the 1980s to open its software platform (the interface of MS-DOS) to other developers.
2
This means that the user does not miss the interface when changing the system.
3
Not to be confused with freeware.
4
http://www.gnu.org
5
This license is called Copyleft.

KAT/ATC UPV/EHU
An introduction to operating systems 9

On another level it should be noted that as Winners and losers

computer technology was occupying new In the early days (50s and 60s of the twentieth
application areas, niche markets for new types of century), the operating system was developed in
operating systems have been developed. A machine language by the manufacturer of the
architecture, which distributed the system as an
remarkable example is the market of real-time indivisible package. The operating system and the
operating systems, for long time common in architecture were absolutely interdependent. Later,
after the experience of Unix and the C programming
industry (control systems), and more recently in language, software and hardware manufacturers
other areas (e.g., video decompression in a specialized, allowing, in principle, the operating
multimedia system). In real-time systems response system to be easily transported to different
platforms (the core of Unix contained 1000 lines of
times are limited by a deadline. After the deadline, machine code, dependent of the architecture). As a
the response is invalid and can even be consequence, the architecture could support
different operating systems. However, the
catastrophic (think about the stability control of a introduction of personal computers made evident
car). Often these types of systems are embedded the need of some form of standardization of
in more complex systems (for example, the control operating systems, from the interface for
applications to the user interface. Standardization
of the stability in a vehicle). Although general came by way of the facts from two factors: the
purpose operating systems (such as Windows, strategic alliance between IBM and Microsoft, and
the opening of the hardware (PC) and software (MS-
Linux or Mac OS) allow running certain DOS interface) platforms to other manufacturers.
noncritical real-time applications (like a video This was at the expense of Apple, the major
cassette recorder, VCR), there are specific real- competitor of Microsoft, which started in the 80s
with an undoubted technological advantage, but
time operating systems (e.g., QNX, FreeRTOS and closed its platform to their own products. As the PC
many others). Many general-purpose operating architecture was conquering markets, Microsoft
systems also support real-time tasks, but are only systems conquered the OS market. The emergence
of Linux and the philosophy of free software in the
suitable when missing the deadline is not critical 90s occurred too late to respond to the
(e.g., multimedia applications). monopolizing inertia of Windows systems. The early
history of the personal computer can be found in
the book Fire in the Valley: The Making of a
At present operating systems, beyond its original Personal Computer, by Paul Freiberger and Michael
orientation, have had to be adapted to a multitude Swaine, brought to the screen by Martyn Burke with
the title The Pirates of Silicon Valley.
of devices, such as mobile phones and other
consumer devices. To this we must add the
embedded systems, increasingly present in our
environment (appliances, cars, industrial plants, robots). Typically embedded systems are subject
to physical constraints and have real time requirements, sometimes critical, leading to specific
solutions, as already mentioned. With respect to the world of mobile devices (smartphones or
tablets), in some cases conventional operating systems have been adapted to the constraints of the
devices (size and power), such as Microsoft Windows Mobile, Apple iPhone OS or Palm OS. In
other cases specific systems have been developed, such as Symbian OS or Google Android.

8 Examples of operating systems

We will discuss here in more detail the history and main characteristics of the more relevant
operating systems, in line with the concepts introduced in the previous sections. We will focus on
those families of operating systems that have made history in computing and whose innovations,
directly or indirectly, remain today.

IBM mainframe operating systems

IBM was for many years the dominant computer company in the market for hardware, operating
systems and applications. His first major operating system, OS/360, whose development ended in
1964, was a complex batch multiprogramming system that stored tasks in partitions (of fixed or
variable size, depending on the version). One version, TSS/360 (Time Shared System, 1967),

KAT/ATC UPV/EHU
An introduction to operating systems 10

offered time-sharing and multiprocessing (with two CPUs), although its enormous complexity (all
systems at that time were developed in assembler) caused that it never worked too well and that its
spread was low.

MVS (Multiple Virtual Storage, 1974) provided virtual memory. It introduced the concept of
virtual machine, which allowed running multiple copies of the operating system into independent
logical partitions, providing a high degree of safety. The MVS architecture has survived and today
is part of the z/OS system.

VMS from Digital

By 1970 the introduction of integrated circuits had cheapened significantly the cost of computers
and expanded its area of use. It appeared the concept of minicomputer to designate a range of
affordable computers (on the order of tens of thousands of Euros) with a small size (like a small
closet). At that time, Digital Equipment Corporation triumphed with its PDP minicomputer family.
The PDP-11, of 16-bit, was the culmination of the saga. It worked with the RSX-11 operating
system, designed to support real-time applications.

The inherent limitation of the 16-bit architecture led Digital to introduce in 1977 the VAX-11
architecture (Virtual Address eXtension), of 32 bits, and the VMS operating system VMS (Virtual
Memory System). One of the features of VMS is its adaptability to the diverse hardware support
level of the different implementations of the VAX architecture, especially regarding virtual
memory. Another feature is that the file system manages file versions, identified by a suffix
denoting the version which is part of the file name. It has a sophisticated process scheduling policy
based on dynamic priorities. Many of the ideas present in VMS were adopted in the development of
Microsoft Windows NT. In 1991 VMS was renamed OpenVMS for the Alpha architecture, the
successor of VAX.

The Unix family

In 1970 at Bell Laboratories of AT&T they started to develop a Unix system, which would have a
great impact and subsequent development. Their ancestors were the CTSS and Multics systems.
The latter, although not commercially successful, set the way for future operating systems. Unix,
whose first version was developed in assembly language on a PDP-7, was entirely rewritten in 1972
in C (language developed at Bell Labs specifically for the Unix project), being the first operating
system written in a high level language. In 1974 there was already a public description of it.

AT&T distributed Unix freely; so many universities and companies used it for their computers and
developments. Since the source code was made public, Unix has a lot of ramifications (Digitals
Ultrix, Microsofts Xenix, IBMs AIX, HPs HP-UX...), but basically there are two families:
System V from AT&T and BSD from the University of Berkeley, whose most popular version was
marketed by Sun Microsystems. While the latter is more powerful with regard to network support,
the two families were unified in the System V Release 4 (SVR4), which in Suns version was called
Solaris.

Unix versions are also available for PCs, being the most popular SCO or Santa Cruz among the
commercial versions, and Linux and FreeBSD among the freely distributed. Linux is a project
initiated by Linus Torvalds in the University of Helsinki in the early 1990s, which proposed free
operating system software in the line of GNU (General Public License) and Free Software
Foundation in the field of applications. Linux is having a huge success not only in small servers,
but also in large machines. Its introduction into the market of personal computers is increasing,

KAT/ATC UPV/EHU
An introduction to operating systems 11

thanks to major advances in three areas: ease of installation, friendly graphical environments, and a
growing number of quality office applications.

Figure 2 shows, in a simplified form, the Unix family tree.

Unix is multiprogrammed, multiuser and multi-terminal, and supports various interfaces both
alphanumeric (shell, C-shell, K-shell...) and graphical (Openwin, Motif, KDE, Gnome...). The latest
versions support even multiprocessing.

Figure 2. The Unix family (Source: wikipedia).

Microsoft: from MS-DOS to Windows NT

When IBM decided in 1980 to enter the world of personal computing, they proposed Microsoft the
development of a new operating system for their new PC. Thus, in August 1981 IBM markets its
first PC with MS-DOS as operating system. MS-DOS 1.0 was compatible with CP/M, the
operating system used by most existing microprocessors until then, but also had significant
improvements on it. It had more information about each file, a better allocation algorithm for disk
space, and was more efficient. However, it could only contain a single directory of files supporting
a maximum of 64 files. It occupied only 8 Kbytes.

KAT/ATC UPV/EHU
An introduction to operating systems 12

When the PC XT appeared (1983), which included a hard drive, Microsoft developed the second
version of MS-DOS, with support for hard disk and hierarchical directories. It also incorporated
some Unix features, such as I/O redirection.

In 1984, with the PC/AT, the Intel 80286 processor offered extended addressing and memory
protection mechanisms. Microsoft introduced the version 3.0 of MS-DOS, which did not take
advantage of the new processor. There were several notable updates in this release. Version 3.1
included network support. From here successive versions of MS-DOS are appearing without major
structural changes.

There are two remarkable facts behind the success of MS-DOS: (a) the appearance, with the
blessing of IBM, of cheap PC clones to which Microsoft provided software Microsoft kept MS-
DOS as proprietary operating system, and (b) maintaining compatibility with previous versions.
The latter resulted, however, in MS-DOS being a less developed system than others from their
competitors.

After IBM choose its own operating system OS/2 for PCs, Microsoft released Windows 3.0 in
1990, copying the idea of the graphical user interface previously marketed by Apple. Windows is
just an interface for MS-DOS and does not provide true multitasking. Still it was a great success
and its use spread rapidly.

Windows 95/98. In 1995 Microsoft had already released Windows NT, a new operating system
designed from scratch for the server market, but the hardware of personal computers of the time
was very limited to support it. Moreover, Windows 3.11 was ridiculously primitive compared to
other less prevalent systems like Mac OS from Apple, which had long offered multitasking,
memory protection and 32-bit addressing. In light of this, Microsoft decided to redesign Windows
3.11 to provide these features, while remaining compatible with 16-bit applications of Windows 3.x
and MS-DOS, marketing it under the name of Windows 95. The Windows 98 and Windows ME
(Millennium Edition) systems are a continuation of Windows 95.

Windows NT/2000/XP/Vista/7. In 1988, Microsoft hired Digital engineers with experience in the
development of VMS, for a new operating system project called Windows NT (New Technology).
The aim is to develop an operating system that integrates new design concepts: client/server
architecture based on a microkernel and multiprocessor support. The microkernel structure was
diluted through successive versions. Early versions from NT 3.1 in 1993, to NT 5.0, traded as
Windows 2000 are aimed at workstations and servers. In 2001 version 5.1 is released, marketed
as Windows XP, which includes for the first time a specific version for home use, ending Windows
95/98 and, thus, the support line of 16-bit applications. Windows XP includes versions for 64-bit
processors. NT 6.0 (Windows Vista), launched in 2007, represents a significant revision of the
architecture, including a new graphical interface and new protection mechanisms, in addition to
many services. This results in a high avidity of resources which obsoletes much of the personal
computer park. The successor is launched in 2009, NT 6.1 (Windows 7), which refines the
implementation to improve performance and also updates forms of user interaction.

Mac OS

In 1979 Xerox PARC gave Apple the rights to use its graphical interface, which included elements
such as icons and mouse. Apple included this interface in the personal computer Lisa (1980), which
pioneered the Macintosh (1984) and the Mac OS operating system. Apart from its advanced
graphical interface, Mac OS offered cooperative multiprogramming (a form of time-sharing in
which each task is responsible for giving the processor to another task). In its early years, the
Macintosh was a huge success, but its relatively high price and closed system strategy motivated

KAT/ATC UPV/EHU
An introduction to operating systems 13

that Microsoft, mainly thanks to its partnership with IBM, imposed its MS-DOS, despite the delay
in introducing a decent graphical interface.

Mac OS evolved to version 9 (1999). In 2000, Apple sells the new Mac OS X, derived from
NeXTSTEP, an operating system based on the Mach 3.0 microkernel. Mac OS X incorporates BSD
Unix code and provides its system call interface. Later Apple adopted Intel as hardware platform, in
substitution of Motorola.

Apple has adapted Mac OS X for mobile devices, marketed under the name iOS. Apples leading
position in this market ensures a good spread of iOS.

Bibliography
A.S. Tanenbaum: Modern Operating Systems (3rd edition). Prentice-Hall, 2008.

W. Stallings: Operating Systems (5th edition). Prentice-Hall, 2004.

Wikipedia: http://en.wikipedia.org

KAT/ATC UPV/EHU