1. Categories of I/O Devices
Human readable
Communication with the user
Printers
Video display terminals
Display
Keyboard
Mouse
Machine readable
Communication with electronic equipment
Disk drives
Sensors
Controllers/Actors
Communication
Communication with remote devices
Digital line drivers
Modems

2. Differences in I/O Devices
Complexity of control
Unit of transfer
Data may be transferred as a stream of bytes for a terminal or in larger blocks for a disk
Data representation
Encoding schemes
Error conditions
Devices respond to errors differently

3. Differences in I/O Devices
Programmed I/O
Process is busy-waiting for the operation to complete
Interrupt-driven I/O
I/O command is issued
Processor continues executing instructions
I/O module sends an interrupt when done
Data Rate
Human input can be very slow
Machine communication should be fast

4. Differences: Data Rate

5. Operating System Design Issues
Efficiency
Most I/O devices extremely slow compared to main memory
Use of multiprogramming allows for some processes to be waiting on I/O while another process executes
I/O cannot keep up with processor speed
Swapping is used to bring in additional Ready processes which is an I/O operation

6. Operating System Design Issues
Generality
Desirable to handle all I/O devices in a uniform manner
Hide most of the details of device I/O in lower-level routines so that processes and upper levels see devices in general terms such as read, write, open, close, lock, unlock

7. Data Transfer Scheme Block-oriented Stream-oriented
Information is stored in fixed sized blocks
Transfers are made a block at a time
Used for disks and tapes
Stream-oriented
Transfer information as a stream of bytes
Used for terminals, printers, communication ports, mouse, and most other devices that are not secondary storage

8. I/O Buffering Reasons for buffering
Processes must wait for I/O to complete before proceeding (you can’t do anything useful anyway)
Adjust speed differences between information source and use
Example: Printer buffer
You don’t want your computer to sit and wait in order to transfer data until there is no other job ahead in the queue and the data could go directly to printer
Instead printer buffers data

9. Techniques for Performing I/O
Direct Memory Access (DMA)
DMA module controls exchange of data between main memory and the I/O device
Processor interrupted only after entire block has been transferred
CPU independent, CPU can do something else while I/O data transfer to memory happens

10. Direct Memory Access
Takes control of the system form the CPU to transfer data to and from memory over the system bus
Cycle stealing is used to transfer data on the system bus
The instruction cycle is suspended so data can be transferred
The CPU pauses one bus cycle
No interrupts occur, no process swapping
Do not save context

11. Communication Inputs I/O = Input + Output Output: CPU initiated
Input: Initiated by device
Creates an interrupt to inform CPU of input
Typical Example: Communication in Client Server environment

12. Examples of Client Server
Application-level protocols provide high-level services
DNS
Electronic mail
TELNET: Remote login
FTP
HTTP World Wide Web
All of these applications use client-server architecture

13. Client Server Architecture
Client Computer
Server Computer
Client
Application
Client Request
Server
Application
Server Response
Network

14. Client Server Paradigm
Server application is ``listener''
Waits for incoming message
Performs service
Returns results
Client application establishes connection
Sends message to server
Waits for return message

15. Issues Identification of Server computer
Identification of Client computer
Identification of Server Application
Identification of Client Application
How do you ensure that both application “understand” each other?
Answer: IP address, Port number , TCP

16. Client Arbitrary application program
Becomes client when network service is needed
Also performs other computations
Invoked directly by user
Runs locally on user's computer
Initiates contact with server
Can access multiple services (one at a time)
Does not require special hardware or sophisticated operating system

17. Server
Special purpose application dedicated to providing network service
Starts at system initialization time
Runs on a remote computer (usually centralized, shared computer)
Waits for service requests from clients; loops to wait for next request
Will accept requests from arbitrary clients; provides one service to each client
Requires powerful hardware and sophisticated operating system

18. Server Class Computer
Shared, centralized computers that run many server applications called ``servers''
More precisely, the applications are the ``servers'' and the computer is a ``server-class computer''
Servers can run on very simple computers...

19. Message exchange Typically, client and server exchange messages:
Client sends request, perhaps with data
Server send response, perhaps with data
Client may send multiple requests; server sends multiple responses
Server may send multiple response - imagine video feed

20. Message Protocols TCP
Standards for messages that ensure that the server can understand the client independent of implementation

21. Multi Server and Client
Multiple clients share same server
One server class computer runs multiple server applications and types (eureka)
Listener organizes

22. Service Identification
Each service gets a unique identifier; both client and server use that identifier
Server registers with local protocol software under the identifier
Client contacts protocol software for session under that identifier
Example - TCP uses protocol port numbers as identifiers
Server registers under port number for service
Client requests session with port number for service
How does client knows the port number on the server computer?

23. Multiple servers for one service
Responding to a client request may require significant time
Other clients must wait while earlier requests are satisfied
Multiple servers can handle requests concurrently, completing shorter requests without waiting for longer requests

24. Master-slave servers
One way to run concurrent servers is to dynamically create server processes for each client
Master server accepts incoming requests and starts slave server for each client
Slave handles subsequent requests from its client
Master server then waits for next request

25. Examples of server running on eureka
Ps –u root
Mountd
In.telnetd
Inetd
Listen:
Program that does the listening for all server and starts a new serving program

26. Review Interrupts
External devices (network card) creates interrupts that are first dealt with by an interrupt controller (IC)
IC sends highest priority interrupt to CPU and puts into the interrupt register the PC associated with the code that can deal with this type of interrupt
Listener decides what to do
Start new server
Start running server who was the recipient of incoming message

27. Selecting from multiple servers
How do incoming messages get delivered to the correct server?
Each transport session has two unique identifiers
(IP address, port number) on server
(IP address, port number) on client
No two clients on one computer can use same source port
Thus, client endpoints are unique, and server computer protocol software can deliver messages to correct server process