Computing Architectures

1
Computing Architectures

• Centralized Computing
• Multiple users access a central computer
  (mainframe/midrange) via remote terminals.
• Central computer performs all the computations
  and controls all the peripherals.
• a.k.a. host-centric computing.
• Greater security is the key advantage.
• Total reliance on the central computer is the key
  disadvantage.

2
Computing Architectures

• Centralized Computing (Contd.)
• Other issues are managing on-line user requests
  and peak-load problems.
• Centralized functions still needed for large
  business operations on-line access to databases
  for airlines, distributions, large
  manufacturers.

3
Computing Architectures

• Personal Computing
• Individual microcomputers with user friendly GUI
  interfaces.
• Greater flexibility for individual users is the
  key advantage.
• Difficulty in sharing the work individuals do and
  the duplication of underutilized hardware and
  software and the key disadvantages.

4
Distributed Computing

• Multiple workstations are linked using a network
  to share data and computing resources.
• Advantages are sharing of data, messages and work
  and sharing of resources.
• Complex to administer security and control
  issues -- access privileges for shared items.

5
Computing Architectures

• Network Computing
• Combines benefits of centralized computing with
  the flexibility and responsiveness of distributed
  computing.
• Multiple network computers (NCs) are linked to a
  central server, which provides links to other
  servers.
• NCs do not contain hard disks.

6
Computing Architectures

• Network Computing (Contd.)
• Greater sharing of work and resources and easier
  to administer than distributed computing.
• Advantages are cheaper network computers,
  centralization of data and programs and greater
  data security.
• Disadvantages are reliance on centralized control
  and the relative immaturity of software designed
  to support this architecture.

7
Client/Server Computing

• Client Server Distributed Computing (Comer
  1994).
• Client/Server computing forms the basis for
  distributed computing.
• Facts - Programs communicate and not computers.
  Computers can run multiple programs at one time.
• A server program offers a service. A client
  program is a consumer of services.

8
Client/Server Computing

• Distinguishing Characteristics of Client/Server
  Computing (Orfali et al. 1999)
• Service - Server is a provider client is a
  consumer of services.
• Shared resources - A server can service many
  clients simultaneously and regulate their access
  to shard resources.

9
Client/Server Computing

• Distinguishing Characteristics of Client/Server
  Computing
• Transparency of location - The server is a
  process or program that can reside on the same
  machine as the client or on a different machine
  across the network. A program can be a client, a
  server, or both.
• Mix-and match - The ideal client/server software
  is independent of hardware or OS platforms.

10
Client/Server Computing

• Distinguishing Characteristics of Client/Server
  Computing
• Message-based exchanges - clients and servers
  interact through a message-passing mechanism.
• Encapsulation of services Message specifies
  what and not how.

11
Client/Server Computing

• Distinguishing Characteristics of Client/Server
  Computing
• Scalability
• Horizontal - add/remove clients
• Vertical - larger and faster server or load
  distribution across multiple servers
• Integrity - Server code and server data are
  centrally managed

12
Types of Servers

• The idea of splitting an application along
  client/server lines results in various forms of
  networked software solutions.
• Each of these solutions is distinguished by the
  nature of service provided.
• File servers, Database servers, Transaction
  servers, GroupWare servers, etc.

13
File Servers

• Client (usually a PC) requests for a file of
  records over a network.
• Many message exchanges over the network to find
  the requested data.
• File servers are key to shared repositories of
  documents, images, engineering drawings, and
  other large data objects.

14
Database Servers

• Client passes SQL requests as messages to the
  database server.
• DBMS code and data reside on the server.
  Application code resides on the client.
• Server performs the search and returns the
  results to the client.
• Play a key role in data warehousing and decision
  support systems.

15
Transaction Servers

• Client invokes remote procedures or services on
  the server.
• Remote procedures execute a group of SQL
  statements called transactions.
• Application is called Online Transaction
  Processing or OLTP.

16
Transaction Servers

• Client side usually includes a GUI server side
  SQL transactions.
• Quick response, high security and integrity
  characterize OLTP application.

17
GroupWare Servers

• Management of semi-structured information such as
  text image, mail, bulletin boards, and the flow
  of work.
• Lotus Notes MS Exchange are examples.
• Communication middleware between the client and
  the server is vendor-specific. E-mail is becoming
  a standard messaging middleware.

18
Fat Clients and Fat Servers

• How is the application split between the client
  and the server?

Source Orfali et al. 1999
19
Fat Clients and Fat Servers

• Fat Client
• More traditional form of client/server.
• Bulk of the application runs on the client.
• Eg. File server model.

20
Fat Clients and Fat Servers

• Fat Servers
• Most of the code runs on the server
  mission-critical applications.
• Network interchanges are minimized.
• Eg. Transaction Server model server encapsulates
  the database client issues procedure calls
  (invokes services).

21
2-Tier, 3-Tier, N-Tier

• Same basic idea as fat servers and fat clients.
• How are the functional units of a client/server
  application assigned either to the client or to
  one or more servers?
• Typical functional units - User Interface,
  Business Logic, and the Shared Database.
• Middleware is used to communicate between the
  tiers.

22
2-Tier, 3-Tier, N-Tier

• In a 2-Tier system, the business or application
  logic is on the client with the GUI or on the
  server with the database or both. E.g., File and
  Database servers.
• In a 3-Tier system, the application logic lives
  in the middle tier. E.g., Web Application server.

23
2-Tier Client/Server Architecture
24
3-Tier Client/Server Architecture
25
Middleware

• A vague term that covers all the distributed
  software needed to support interactions between
  clients and servers. (Orfali et al. 1999)
• Performs whatever translation is necessary to
  make a clients request understandable to a
  server program. (Alter 1999)

26
Middleware

• Starts on the client side that invokes a service,
  covers the transmission of the request over the
  network and the resulting response.
• Middleware does not include either the software
  that provides the actual service or the database.
• On the client side, it does not include the
  interface.

27
Middleware (Contd.)

• It does include software used to co-ordinate
  inter-server interactions.
• Represents a rapidly changing and growing area.
  (From 1.7 billion in 1997 to 7 billion in
  2002 according to IDC.)

28
2-Tier Client/Server

• Simplicity is the biggest selling point.
• Well suited for departmental applications (DSS,
  small-scale GroupWare) or simple Web publishing
  applications.
• Scalability problems - hard to deploy and manage
  fat clients on a large scale.
• Difficulty in managing software on remote
  desktops.

29
3-Tier (N-Tier) Client/Server

• For large-scale, enterprise-wide applications
  that serve thousands of enterprise clients.
• Applications are split into hundreds of
  components and distributed across multiple
  servers.

30
3-Tier (N-Tier) Client/Server

• In theory, 3-Tier systems are
• scalable - load distribution across multiple
  servers.
• flexible - all 3 tiers on different computers
  second and third on the same computer second
  tier across multiple servers etc.
• More secure - data tables are encapsulated.

31
3-Tier (N-Tier) Client/Server

• In theory, 3-Tier systems are
• Easier to administer - less logic on the client
  to manage.
• More responsive - only service requests and
  responses (and not raw data) are sent between the
  client and the server.

32
3-Tier Vs. N-Tier

• Middle tier in most 3-Tier applications is often
  implemented as a collection of components.
• Components are used in a variety of
  client-initiated business transactions.
• Components co-operate while responding to a
  client request.
• Components can encapsulate legacy applications
  running on main frames.

33
Component-based Architecture

• Advantages of component-based applications in the
  middle tier.
• Lend themselves well to a small-team, incremental
  development approach.
• Components can be reused.
• Components encapsulate details of application
  logic.

34
Component-based Architecture

• Advantages of component-based applications in the
  middle tier.
• Clients can access data and functions easily and
  safely.
• Possibility of buying ready-made components.
• Basis for suites of applications.

35
Lecture 6 Computing Architectures

• References
• Alter S. (1999), Information Systems A
  Management Perspective, Third Edition,
  Prentice-Hall, Inc., New Jersey.
• Chapter 8 pp. 268 - 276
• Comer, D.E. (1997), The Internet Book, Second
  Edition, Prentice-Hall, Inc., New Jersey.
• Chapter 15 pp. 121 - 125
• Orfali R., D. Harkey and J. Edwards (1999),
  Client/Server Survival Guide, Third Edition, John
  Wiley.
• Chapters 1-3