Chapter 16: Distributed Processing, Client Server, and Clusters

1
Chapter 16 Distributed Processing, Client /
Server, and Clusters

• CS 472 Operating Systems
• Indiana University Purdue University Fort Wayne

2
Brief review of Chapter 17 Networking

• Note Chapter 17 is an online chapter and does
  not appear in the textbook
• Available under Online Chapters at
• WilliamStallings.com/OS/OSe6.html
• Be aware that the URL is case sensitive
• A .pdf of this chapter should also be available
  on the class web site under Resources

3
Distributed System Architectures

• There is a spectrum of distributed system
  architectures
• Communications architecture
• Network OS
• Distributed OS

4
Communications system architecture

• TCP/IP is the most widely used example
• Software supports a group of networked computers
• Appropriate for email and file transfer
• Separate OS on each computer

5
TCP/IP review

• TCP/IP is a protocol suite
• Vendor independent
• DoD standardization (originally for ARPANET)
• Five relatively independent layers
• Physical
• Network access
• Internet
• Transport (host-to-host)
• Application

6
TCP/IP review

• Transport layer ensures all data arrives at the
  destination and in the order sent
• Protocols
• IP (Internet Protocol) internet layer
• TCP (Transmission Protocol) transport layer
• SMTP (Simple Mail Transfer Protocol)
  application layer
• FTP (File Transfer Protocol) application layer
• TELNET (remote login, scroll-mode) application
  layer

7
Network OS system architecture

• One or more servers
• These provide network services such as file
  storage
• Separate OS on each computer
• Network OS simply allows interaction with the
  servers
• Example
• Can map a foreign disk drive Z on a server to
  local PC
• Foreign drive appears under My Computer as a
  local drive ( like C )

8
Distributed OS system architecture

• Common shared OS on all computers on the network
• Permits a process to migrate to another host
  computer
• Now
  on to Chapter 16 . . .

9
Client / server computing

• Client machines are usually single-user PCs or
  workstations that provide a user-friendly
  interface to the end user
• Each server provides a set of shared services to
  the clients
• The server enables many clients to share access
  to the same database and enables the use of a
  high-performance computer system to manage the
  database

10
Client/Server Terminology

• Applications Programming Interface (API)
• A set of functions and call programs that allow
  clients and servers to intercommunicate
• Client
• A networked information requester, usually a PC
  or workstation, that can query database and/or
  other information from a server
• Middleware
• A set of drivers, APIs, or other software that
  improves connectivity between a client
  application and a server
• Relational Database
• A database in which information access is limited
  to the selection of rows that satisfy all search
  criteria
• Server
• A computer, usually a high-powered workstation, a
  minicomputer, or a mainframe, that houses
  information for manipulation by networked clients
• Structured Query Language (SQL)
• A language developed by IBM and standardized by
  ANSI for addressing, creating, updating, or
  querying relational databases

11
(No Transcript)
12
Server types

• Disk server
• Provides shared drive like Z
• File server
• Provides logical structure, protection,
  concurrent access, etc.
• Print server
• Provides SPOOLing
• Simultaneous Peripheral Operation On-Line

13
Server types

• Database server
• Searches for and returns a small number of
  records
• Server software is typically SQL
• Client does data analysis with application
  software
• Server can do more work if analysis involves a
  large number of records
• It is not efficient to send 1,000,000 records
  over a network
• This requires some application-specific software
  to reside on the server

14
Classes of Client/Server Applications

• Host-based processing
• Not true client/server computing
• Traditional mainframe environment

dumb terminal
15
Classes of Client/Server Applications

• Server-based processing
• Server does all the processing
• Client provides a graphical user interface

16
Classes of Client/Server Applications

• Cooperative processing
• Application processing is performed in an
  optimized fashion
• Complex to set up and maintain

17
Classes of Client/Server Applications

• Client-based processing
• All application processing done at the client
• Data validation routines and other database logic
  functions are done at the server

18
Three-Tier Client/Server Architecture

• Increasingly, this architecture is used
• Application software distributed among three
  types of machines
• User machine
• Thin client
• Middle-tier server
• Gateway
• Conversion protocols
• Merge/integrate results from different data
  sources
• Backend server

19
(No Transcript)
20
File Cache Consistency

• File caches hold recently accessed file records
  on servers and each client
• Various caches must be kept consistent
• More complicated than readers/writers problem
• Writer must write back to disk and server must
  notify all readers to not use cache copy

21
Distributed Message Passing

• Message passing must be used for interprocess
  communication and synchronization
• This is because no shared memory exists
• Therefore, no semaphores
• Two models
• Client-server
• Remote Procedure Call (RPC)

22
Client-server model

• Involves sending and receiving messages using
  primitives as in a single system
• Send( destination, message)
• Receive( from, messageBuffer)
• Note the from parameter may include all
• The primitives rely on a communications
  distributed system architecture like TCP / IP

23
Reliable v. unreliable message passing

• Reliable
• Guarantees delivery if possible
• Not necessary to let the sending process know
  that the message was delivered
• High overhead
• Unreliable
• Sends the message out to the network without
  reporting success or failure
• Reduces complexity and overhead
• Destination may need/want to send acknowledgement

24
Blocking v. nonblocking

• Nonblocking
• Process is not suspended as a result of issuing a
  Send or Receive
• Sender released after a copy of the message is
  made
• Efficient and flexible
• Difficult to debug
• Blocking
• Send does not return control to the sending
  process until the message has been transmitted
  (reliable case)
• OR does not return control until an
  acknowledgment is received (unreliable case)
• Receive does not return until a message has been
  placed in the allocated buffer

25
Remote procedure call

• Allows programs on different machines to interact
  using simple procedure call and return semantics
• Widely accepted
• Standardized
• Client and server modules can be moved among
  computers and operating systems easily

26
Remote procedure call

• Sender has a dummy procedure that sends
  parameters to the target as a message
• The target reacts to the message by calling the
  actual procedure
• Results are then sent back as a message and
  returned by the dummy procedure

27
(No Transcript)
28
Two models of RPC

• Synchronous RPC
• Special case of reliable, blocking message
  passing
• Behaves much like a subroutine call
• Asynchronous RPC
• Does not block the caller
• Enables client execution to proceed locally in
  parallel with server invocation
• Exploits possibility of parallelism
• Example of client server synchronization
• The client can send a sequence of asynchronous
  RPCs followed by a final synchronous RPC
• The server responds to the synchronous RPC only
  after all preceding work has been completed

29
Object-oriented mechanisms

• Clients and servers ship messages back and forth
  between objects
• Common Object Request Broker Architecture (CORBA)
• A client sends a request to an object broker
• The broker calls the appropriate object and
  passes along any relevant data

30
Clusters

• Alternative to symmetric multiprocessing (SMP)
• Group of networked computers, called nodes
• Computers work together as a unified computing
  resource
• Illusion of being one computer
• Provides load balancing among all nodes

31
Advantages of clusters over SMP

• Scalability
• Both incremental and absolute
• Easy to add a new node
• Very large clusters of multiprocessor systems
  possible
• Availability
• Failure of one node does bring system down
• Easy to make components redundant
• Failover is a switch to an alternate node in
  case of failure
• Price / performance
• Building blocks are commodity components

32
Cluster configurations

• Separate server
• Each node has private (non-shared) disk
• Failover capability requires constant copying of
  data to disks on other nodes
• Shared nothing
• Reduces communication overhead
• Common RAID disk, partitioned into volumes
• Each volume is owned by a separate node
• Failure of node only requires transfer of
  ownership of its volume
• Shared disk
• Each node may access all volumes
• Mutual exclusion must be enforced

33
Advantages of SMP over clusters

• SMP is easier to manage and configure
• SMP takes up less space and draws less power
• SMP products are well established and stable