1

ARCHITETCTURES ofCOMPUTERS SYSTEMS

2

Centralized systems

Characteristics

•Proprietary systems

• Limited knowledge of the computer thecnologies

• Backlog

3

TERMINAL

TERMINAL TERMINAL

TERMINAL

centralizedcomputer

Centralized solution

4

Distributed Systems

Set of independent computers connected by acommunication network in order to executedifferent functions (administration,management, thecnical problems, logistic,production..) that are present in a complexorganization.

5

computer

Data

baseLocal terminals

6

Advances in thechnology

• personal computer

• local network (LAN)

• High speed networks (LAN, MAN, WAN)

7

Advantages of Distributed systems over Centralized ones

• servers offer a better price/performance than mainframes

• A large number of applications are distributed ( they run ondifferent machines)

• Reliability

• Incremental growth (computing power can be added in smallincrements)

• Data sharing (allow many users access to a common database)

• Device sharing (allow many users to share expensiveperipherals)

• Flexibility (spread the workload over the available machinesin the most cost effective way)

8

Some definitions

• Reliability: the ability of a system to perform and maintain itsfunctions in routine circumstances, as well as hostile orunexpected circumstances.

• availability: The ratio of (a) the total time a funcional unit iscapable of being used during a given interval to (b) the lengthof the interval

• security: protection against non authorized access

• fault tolerance: the ability of a system to respond gracefullyto an unexpected hardware or software failure

9

Open Systems

• Distributed systems consisting of eterogenous hardwareand software components from different system vendors.

• Unlike a propritary solution a open distributed systemcan be realized by using components from the differentvendors.

• Utilization of standards (hardware, software)

10

Open systems are computer systems that provide somecombination of interoperability, portability, and open standardsoftware.

The term was popularized in the early 1980s, mainly todescribe systems based on UNIX, especially in contrast to themore complex mainframes and minicomputers in use at thattime.

Unlike older legacy system, the newer generation of Unixsystems featured standardized programming interfaces andperipheral interconnects; third party development of hardwareand software was encouraged.

A significant departure from the norm of the time, which sawcompanies such as Amdhal and Hitachi going to court for theright to sell systems and peripherals that were compatible withIBM's mainframes.

11

Standard

• A technical standard is an established norm orrequirement. It is usually a formal document thatestablishes uniform engineering or technical criteria,methods, processes and practices.

• Standard de jure: Hardware or software that isendorsed by a standards organization.

• Standard de facto: Hardware or software that iswidely used, but not endorsed by a standardsorganization

• The standard concept is relative to the functionalaspects of a component; the component may berealized in different ways.

12

CLIENT-SERVER MODEL

Coordination model in a distributed system.

It defines:

• Which process may begin the interaction

• Which process may answer

• How error conditions may be managed.

13

• Although an Internet system provides a basiccommunication service, the protocol software cannot initiatecontrol with, or accept contact from, a remote computer.

• Of course, two application involved in a communicationcannot both wait for a message to arrive. One applicationmust actively initiate interaction while the other applicationpassively waits.

• Most network applications use a form of communicationknown as the client –server paradigm. A serverapplication waits passively for contact, while a clientapplication initiates communication actively.

14

Process classification

• Client process, the process that requires a service

• Server process, the process that provides the requiredservice

• The client requires a service, the server provides the serviceand makes available the results to the client

15

Client functionsIn generally, client software:

• is an arbitrary application program that becames a clienttemporarily when remote access is needed, but alsoperforms other computation locally.

• is invoked locally by a user, and executes only for onesession

• runs locally on a user personal computer

• actively initiates contact with a server

• can access multiple services as needed, but activelycontacts one remote server at a time.

• does not require special hardware or a sophisticatedoperating system

16

Server functions

• Is a special purpose, privileged program dedicated toproviding one service, but can handle multiple remoteclients at the same time.

• Run on a shared computer(i.e. not a user’s personalcomputer).

• Wait passively for contact from arbitrary remote clients

• Accepts contact from arbitrary clients, but offers a singleservice

• Requires powerfull hardware and a sophisticated operatingsystem

17

Application server

Kinds of client services:

• mail server

• file server

• terminal server

• name server

• authentication server

• gateway server

• administration server• ………….

18

A server must guarantee:

authentication: client identity verification

authorization: verification of the possibility for a client toaccess to a particular service

data security: garantee that specific data cannot be readand/or modified

19

SERVER E SERVER CLASS COMPUTERS

• Formally the term server refers to a program that waitspassively for communication and not to the computer on wich itexecutes.

• However,when a computer is dedicated to running one or moreserver programs, the computer itself is sometimes called aserver.

• Server class computer refers to a powerfull computer used torun server software.

• It may offer simultaneosly more services, each of themcorresponding to a specific program application

20

• Information can flow in either or both directions between aclient and a server.

• Typically, a client sends a request to a server, and theserver returns a response to the client.

• In some cases, a client sends a series of requests and theserver issues a series of responses (e.g., a data base servercan allow to a user to look up more than one item at a time)

• In other cases, the server provides continous outputwithout any request . As soon as the client contacts theserver, the server begins sending data (e.g. local weatherserver might send continous weather reports with updatedtemparature and barometric pressure.)

21

Characteristics of a client-server architecture

• Client and server machines need different amount of hardwareand software resources.

• Client and server machines may belong to different vendors.

• orizzontal scalability (increase of the client machines) andvertical scalability (migration to a more powerful server or to amultiserver solution)

• A client or server application interacts directly with a transportlayer protocol to establish communication and to send or receiveinformation.

• The transport protocol then uses lower layer protocols to send orreceive individual messages. Thua, a computer needs a completestack of protocols to run either a client or a server.

22

• A single server-class computer can offer multiple servicesat the same time; a separate server program is needed foreach service.

•Identifyng a particular serviceTCP uses 16-bit integer values (protocol port numbers) toidentify services, and assign a unique port number to eachservice.

•A client specifies the protocol port number of the desiredservice when sending a request.

23

Classification of client server structures

• In a client/server application three functions are present:

• user interface• application programs• data management

• Following the assignment of functions among client andserver we have three possible types of structures:

- host-based processing- server based processing- cooperative processing- client based processing

24

client server

Host-based processing

Server-based processing

Presentation logic

Application logic

Database logic

DBMS

Presentation logicApplication logic

Database logic

DBMS

25

• Host- based processing is not true client server computing.It refers to the traditional mainframe environment in which allor virtually all of the processing is done on a central host.The user’s station is generally limited to the role of a terminalemulator.

Server-based processing. The client is principallyresponsible for providing a graphical user interface, whilevirtually all the processing is done on the server.

26

client server

Cooperative processing

Client-based processing

Application logic

Database logic

DBMS

Presentation logic

Database logic

DBMS

Presentation logic

Application logic

Application logic

Database Logic

27

Client-based processing. Virtually all applicationprocessing may be done at the client, with the exception ofdatabase logic functions that are best performed at theserver.This configuration is perahps the most common clientserver approach in current use.

Cooperative processing. That application processing isperformed in an optimized fashion, taking advantage of thestrenghts of both client and server machines and ofdistribution of data.

28

Three-level Client/Server Architecture

The traditional client/server architecture involves twolevels, a client level and a server level.

A three level architecture is constituted by three types ofmachines: a user machine, a middle-level server and a backend server.

The user machine (client) is typically a thin client.

The middle-level server is the application server.

The back-end server is the data server.

29

Legacy system

• A “hold” system (mainframe) which has not beenremoved after the installation of a new system.

30

Dynamic server creation

• Concurrency is fundamental to the client server model ofinteraction because a concurrent server offers service tomultiple clients at the same time, without requiring eachclient to wait for previous clients to finish.

•The server creates a new thread for each request thatarrives.

•The main server thread waits for a request to arrive, then itcreates a new service thread to handle each client

31

Complex Client-Server Interactions

• A client application is not restricted to accessing asingle service. The client may contact a different server(perhaps on a different computer) for each service.

•. A server for a service can became a client of anotherserver.

32

STATELESS Vs STATEFULL SERVERS

•State information: information that a server maintainsabout the interaction state with the clients.

•The information state provides an incremental answer to anew request.

•The information state can increase the system efficiency,but problem may arise in the case of duplicated messages ,delays and messages arrived out of order.

33

• File server - remote access by clients . Two kinds ofrequests : Reading data from the file and writing dataon the file.

• Stateless server – It does not mantain informationsabout the transactions. Each request must specify thename of the file, the position in the file from which toextract or to insert information, the number of bytesto insert or extract.

• Statefull server – The server mantains the stateinformation about the files that accessed the server

34

Statefull server

The server sends to the client the identifier of the previouslyopen file. The client utilizes this identifier for the followingrequests.

identifier file name currentposition1 test.program.c 02 tcp.book.doc 4563 dept.budget.txt 384 tetris.exe 128

35

Statefull server

• Advantages: higher efficiency, incrementaloperations, lower dimension messages.

• Problems: comand duplication, delays, messagesout of order, server crash (open files are notcorrectly closed).

36

COOKIES

•Information records that the server sends to the clienttogether with the service answer.

•A cookie containsuser informationsinformations on tha visited pages

•The following requests of the client to the server requirethe transmission of the relative cookies.

37

Cookies

The server answer to a request of a client contains:

Set-cookie: user identification number created by the server web

Set-cookie: 1678453.

The client inserts this information in a cookie file . Theinformation includes the host name and the identification numberof the client.

In the following requests the client inserts the previousinformation .

The server does not know the user name, but it is able toidentify the client.

38

The web servers use cookies for the followingobjectives:

To eliminate the need to require user name andpassword in order to authenticate the client.

To store the user preferences

To keep track of the objects that a client isbuying on the web site.