SCHOOL OF ENGINEERING

PROFESSIONAL SCHOOL OF SYSTEMS AND COMPUTER ENGINEERING

Topic: Distributed Operating System Sprite

PRESENTED BY:
Alumnos : -Wilfredo Carrión Umeres
José A. ChahuaTaípe
Yamil N. Solís Diaz
Instructor: Eng. Javier Moreano Córdova
Curso:Sistemas Distribuidos

Abancay - Apurímac
Peru
 WHAT IS A DISTRIBUTED OPERATING SYSTEM?

Distributed Operating Systems is one that facilitates access and management of the
distributed resources on the network. It is where users can access remote resources from it.
way they do it for local resources allowing the distribution of jobs, tasks, or processes,
among a set of processors. It may be that this set of processors is in a machine or
in different ways, which is transparent to the user.
 DISTRIBUTED OPERATING SYSTEM: SPRITE

• Sprite is the name of a distributed operating system with a monolithic kernel developed by the University of

California, Berkeley, more specifically by the research group of John Ousterhout.

• This distributed operating system has a monolithic kernel, as the distribution occurs within the kernel itself.

• Type of program:Distributed Operating System

• Developer:University of California, Berkeley
• Type of nucleus:Monolithic

• Supported Platform Types: DEC Station 5000 Mod 200 and SPARCstation 2

• Latest Version: 1992

 SPRITE

• This operating system has the appearance for programmers of a unique system, as the
distribution occurs within the core itself and in this way.

• Sprite gives us the impression of working on a typical UNIX system.

• Sprite will work as if it were a large simple time-sharing system or an image of a

simple system.

• It is possible to run processes on other workstations explicitly, using a

remote connection to the workstation where it is desired that the processes run.
 DESIGN

• The Sprite design team adopts a traditional computing model,

following the UNIX line, with network-connected workstations. They support
that the distributed nature of the system should remain hidden in the core.
• In the way that processors are associated with processes, the Sprite OS for each user
it has its own workstation where it runs its processes, in addition it provides a
migration mechanism for transferring processes to other machines that are not being
used.
• Sprite uses migration of processes transparently to utilize inactive nodes. The design of Sprite aims at two main objectives:

• the load distribution must be transparent to the user, and when an inactive node returns to activity due to the creation of
new processes specific to that node, the processes migrated to that node must immediately return to the origin node.

• These expelled processes have no residual dependency on the node from which they were expelled.

• The implementation of process migration at the kernel level and load distribution at the user level. Sprite uses the concept
from the source node, which is the node where the process would always have to be running if the migration did not exist.

• This operating system uses two ways to make system calls. Those system calls that are node independent.
where it executes, are made on the node where it is running, and those that depend on where they are executed are executed remotely
At the source node. Sprite provides remote access to devices and files, and uses unique process identifiers for the entire network.

• Each node of the network maintains a daemon that monitors the load of that node. When a node is underutilized, it reports that state to
a node called 'central migration server', which is responsible for maintaining the list of underutilized or inactive nodes.
 HOW IS THE ASSIGNMENT OF PROCESSES GIVEN TO
PROCESSORS
• Sprite also allows the automatic and transparent execution of processes in others
workstations, using the procedure known as migration of
processes. Through this mechanism, the workstations that do not
they are being used, a centralized 'daemon' process called 'migd keeps' tracks
the idle servers and assigns them tasks when necessary.
• When the owner of a workstation returns to use it, then the
the process is migrated again, back to the source workstation or to another one that is
unemployed.
 HOW APPLICATIONS SEE SERVICES
THE SPRITE
• Use cache for files on workstations and also on the server
files, for what consumes space in its main memory, reducing the
transfers on the network and the use of disk bandwidth, achieving high performance
especially with applications that make intensive use of files. The communication
between processes is small, this characteristic is typical of UNIX processes.
• The method for location transparency at the user level is relatively
inefficient. Aspects such as process migration and others have been affected and
limited by compatibility with UNIX.
 DISCONTINUATION

Sprite was not a microkernel system, and it suffered the same

a range of problems that other Unix systems in terms of
development complexity, becoming increasingly difficult
you develop as more functionality was added. In the
1990s, I was suffering, and the small team that supported the
project simply was not able to keep up with
the rapid changes in Unix that were happening at that time
Time. The project was slowly closed in 1994.
 CHARACTERISTICS

• Based on Unix

• It offers compatibility with UNIX applications.

• Use cache for files on workstations

• It is built on a network file system

• It will work in a 'conscious' way online, while being transparent to the user.
 HOW IT WORKS
• Each user has their own workstation where they run their processes.

• Provide a migration mechanism to transfer the processes to other machines that do not
are being used.
 HOW IT WORKS

• Processes run on the workstation where the user launched the application.

• It is possible to run the processes on other workstations.

• It allows for the automatic and transparent execution of processes.

• It tracks the idle servers and assigns processes to them when necessary.

• It measures the time it takes to open and close a file; client cache is considered both enabled and disabled.
REQUIREMENTS: DESKTOP COMPUTER
(DOMESTIC)

• RAM Memory: 512 MB.

• Hard Disk: 5 GB (for a complete installation including swap).
• VGA graphics card and monitor capable of supporting a resolution of 1024x768.
• CD-ROM reader or network card.
• Sound card.
• Internet connection.
 REQUIREMENTS: SERVER

• Processor: 300 MHz x86.

• RAM Memory: 128 MB.
• Disco Duro: 1 GB
• VGA graphics card and monitor
capable of supporting a resolution of 640x480.
• CD-ROM reader
 ADVANTAGES

• One of the advantages of distributed systems in the economy is that it is much cheaper to add servers and clients when they
requires increasing the processing power.
• Working together, for example: in an assembly factory, the robots have their own CPUs and perform actions in
set, directed by a distributed system.
• They have greater reliability. Since the workload is distributed across many machines, the failure of one of them does not affect the others.
besides, the system survives as a whole.
• Shared data: A distributed system allows multiple users to access a common database.
• Shared devices: Similarly, peripherals can be shared among various users, such as a printer.
• Comunicación: un sistema distribuido facilita la comunicación entre computadoras aisladas con el e-mail, por ejemplo.
• Flexibility: A distributed system spreads the workload among the available machines in the most efficient way regarding
the costs
 DISADVANTAGES

• The main problem is the software, it is the design, implementation, and use of distributed software, for
presents numerous inconveniences. The main questions are as follows:
• Computers in a system can be of different types and run different versions of
operating systems
• The system can be accessed from several different computers, and the traffic on the network may be
subject to unwanted eavesdropping (this makes it more difficult to maintain the integrity of the data in the system and
that services do not degrade due to attacks.
• System synchronization: difficult and constant
• Standardization of the system: complicated and limiting
 CONCLUSIONS

The future of Distributed operating systems as well as the Sprite operating system
they are in the formation of a new universal kernel that supports distribution for it
can be applied to all operating systems regardless of their platform. Or for that
less than the systems can be distributed among the computers running this
same system within the same network and unified by an operating system server
distributed. The distributed operating system is often used as subsystems
operations using their advantages such as the cluster system for
storage. We believe that if we can find sub-solutions to the distribution that
If beneficial, we should delve into the future of computing into total distribution.
THANK YOU