tdc561 network programming week 4: client/server design approaches case study using sockets – a...

TDC561 Network Programming

Week 4: Client/Server Design Approaches

Case Study Using Sockets – A Client Server Application of Network Performance Management

presentation prepared with the help of Dale Knudson PhD, Lucent Technologies

Network Programming (TDC561) Winter 2003

References

Douglas Comer, David Stevens, Internetworking with TCP/IP : Client-Server Programming, Volume III (BSD Unix and ANSI C), 2nd edition, 1996 (ISBN 0-13-260969-X) – Chap. 2, 8

W. Richard Stevens, Network Programming : Networking API: Sockets and XTI, Volume 1, 2nd edition, 1998 (ISBN 0-13-490012-X) – Chap. 7 (partial), 27 (partial), 16.1-16.3 (partial)


Server Design

IterativeIterativeConnectionlessConnectionless

IterativeIterativeConnection-OrientedConnection-Oriented

ConcurrentConcurrentConnection-OrientedConnection-Oriented

ConcurrentConcurrentConnectionlessConnectionless


Concurrent vs. Iterative

Iterative

•Small, fixed size requests•Easy to program

Iterative

•Small, fixed size requests•Easy to program

Concurrent

•Large or variable size requests•Harder to program•Typically uses more system resources

Concurrent

•Large or variable size requests•Harder to program•Typically uses more system resources


Connectionless vs. Connection-Oriented

Connection-Oriented

•Easy to program•Transport protocol handles the tough stuff.•Requires separate socket for each connection.

Connection-Oriented

•Easy to program•Transport protocol handles the tough stuff.•Requires separate socket for each connection.

Connectionless

•Less overhead•No limitation on number of clients

Connectionless

•Less overhead•No limitation on number of clients


Statelessness

State: Information that a server maintains about the status of ongoing client interactions.

Issues with Statefullness Issues with Statefullness – Clients can go down at any time.Clients can go down at any time.

– Client hosts can reboot many times.Client hosts can reboot many times.

– The network can lose messages.The network can lose messages.

– The network can duplicate messages.The network can duplicate messages.

ExampleExample– Connectionless servers that keep state information must be

designed carefully

» Messages can be duplicatedMessages can be duplicated


Concurrent Server Design Alternatives

One child per client

Single Process Concurrency (AKA Pseudo-Concurrency or

Apparent Concurrency)

Pre-forking multiple processes

Spawn one thread per client

Pre-threaded Server

will be discussed in a future lecture


One child per client

Traditional Unix server:

– TCP: after call to accept(), call fork().

– UDP: after recvfrom(), call fork().

– Each process needs only a few sockets.

– Small requests can be serviced in a small amount of time.

Parent process needs to clean up after children (call wait() ).


Discussion Stevens Example: Concurrency using fork() 1/3

/* Code fragment that uses fork() and signal()

to implement concurrency */

/* include and define statements section */

void signal_handler(int sig) {

int status;

wait(&status); /* awaits child process to exit

therefore allows child to terminate,

and to transit from ZOMBIE to

NORMAL TEMINATION (END) state

*/

signal(SIGCHLD,&signal_handler);

/* restarts signal handler */

}



main(int argc, char *argv[]){/* Variable declaration section *//* The calls socket(), bind(), and listen() */

signal(SIGCHLD,&signal_handler);while(1) { /* infinite accept() loop */newfd = accept(sockfd,(struct sockaddr *)&theiraddr,&sinsize);if (newfd < 0) { /* error in accept() */if (errno = EINTR)

continue;else {

perror("accept");exit(-1);

}}

See previous slide



/* successfully accepted a new client connection newfd >=0 */

switch (fork()) {

case -1: /* fork() error */

perror("fork");

exit(-1);

case 0: /* child handles request */

close(sockfd);

/* read msg and form a response */

/* send response back to the client */

close(newfd);

exit(-1); /* exit() sends by default SIGCHLD to parent */

default: /* parent returns to wait for another request */

close(newfd);

} /* end switch */

} /* end while(1) */

}


Discussion: Single Process Concurrency 1/2

main(int argc, char *argv[]){/* various declarations */int result;fd_set readfds, testfds;/* socket(), bind(), listen() … *//* inititialization of the fd setFD_ZERO(&readfds);/* add listening socket sockfd to the reading set */FD_SET(sockfd,&readfds);while(1) {int fd;testfds = readfds;result = select(FD_SETSIZE, &testfds, NULL, NULL, NULL);/* error check for select */


Discussion: Single Process Concurrency 2/2

for (fd = 0; fd < FD_SETSIZE; fd++) {

if (FD_ISSET(fd,&testfds)) { /* find activated socket fd */

if (fd == sockfd) { /* ==> New connection */

newfd = accept(sockfd,(struct sockaddr *)&theiraddr,&sinsize);

FD_SET(newfd,&readfds); /* update file descriptor set with newfd */

}

else { /* ==> request from ``old'' connection */

/* call read() and handle request

from old connection using fd */

close(fd);

FD_CLR(fd,&readfds); /* remove fd when connection finished */

} /* else */

} /* end if (FD_ISSET() */

} / * end for (; ; ) */

} /* end while(1) */

return 1;

}


OAM&P - Operations, Administration, Maintenance and Provisioning

ITU Categories

(International Telecommunications Union)

1. Configuration Management (CM)

2. Accounting Management (AM)

3. Fault Management (FM)

4. Performance Management (PM)

5. Security Management (SM)


Performance Management

Wireless SwitchOSS

NM

PM report

(one per hour)

ftp

What if you want the data more frequently?

Custom

Reports

OSS NM – Operation Support System - Network Manager


OMP Server

Every 5 minutes

(but how much data?)

Collection

Specification OMP

Part A – every 1 hour

Part B – every 30 minutes

Part C – every 5 minutes

OMP Server (Operations and Maintenance Platform Server)


PM

Poll for data

PMPM

PM

OMP

Master Clock


OMP

Web browser

Ethernet


OMP

Web browser

Ethernet

Web Server


OMP

Web browser

Ethernet

Web Server

HTML

pages


OMP

Web browser

Ethernet

Web Server

HTML

pages

PM hourly

data

CGI Script

Generated

HTML


OMP

Web browser

Ethernet

Web Server

HTML

pages

PM hourly

data

CGI Script

Generated

HTML

PM On

Demand data


OMP

Web browser

Ethernet

Web Server

HTML

pages

PM hourly data

CGI Script

Generated

HTML

On Demand data

PM

Hourly

Collection

PM

On Demand

Collection


OMP

User Terminal

Ethernet

Web Server

HTML

pages

PM hourly data

CGI Script

Generated

HTML

On Demand data

PM

Hourly

Collection

PM

On Demand

Collection

Web Browser

Java applet


OMP

User Terminal

Ethernet

Web Server

HTML

pages

PM hourly data

CGI Script

Generated

HTML

On Demand data

PM

Hourly

Collection

PM

On Demand

Collection

Server

Web Browser

Java applet

Client

Socket


On Demand

Collection

(server socket)

CREATE A SERVER SOCKET

AF_INET = use TCP/IP internet protocols

SOCK_STREAM = connection oriented socket (not datagrams)

s_sock = socket( AF_INET, SOCK_STREAM, 0)

Note that this socket is not yet associated with a

port number.


On Demand

Collection

(server socket)


s_sock = socket( AF_INET, SOCK_STREAM,0)

bind will associate s_sock with a local name or address/port pair.

memset(&serv_adr, 0, sizeof(serv_adr));

serv_adr.sin_family = AF_INET;

serv_adr.sin_addr.s_addr = INADDR_ANY;

serv_adr.sin_port = htons(PORT);

bind (s_sock, (struct sockaddr *)(&serv_adr), sizeof(serv_adr));

Note that this port number will be known to all the clients.


On Demand

Collection

(server socket)




Set up a queue for incoming connection requests.

In this case set up for a maximum of 5 clients.

listen (s_sock, 5);


Java applet

(socket client)

CREATE A CLIENT SOCKET

AF_INET = use TCP/IP internet protocols

SOCK_STREAM = connection oriented socket (not datagrams)

c_sock = socket( AF_INET, SOCK_STREAM, 0)

Note that this socket is not yet associated with a

local or destination address.


Java applet

(socket client)


c_sock = socket( AF_INET, SOCK_STREAM,0);

Connect binds a permanent destination to a socket, placing it in the connected state.

Internet address interpretation routine – translates from dotted decimal format (e.g., a.b.c.d) to a 32 bit internet address.

serv_adr.sin_family = AF_INET;

serv_adr.sin_addr.s_addr = inet_addr(serv_address);

serv_adr.sin_port = htons(PORT);

connect (c_sock, (*sockaddr *)&server, sizeof(server));


Java applet

(socket client)


c_sock = socket( AF_INET, SOCK_STREAM,0);

connect (c_sock, (*sockaddr *)&server, sizeof(server));

The client can now start reading and writing on the socket.

Do {

buf = message to send to the server

write (c_sock, buf, len);

read (c_sock, buf, len);

process response message received from the server

} while (1); /* or until complete */


On Demand

Collection

(socket server)

Java applet

(socket client)

Known port

number


On Demand

Collection

(server socket)




listen (s_sock, 5);

Once a socket has been set up, the server needs to wait for a connection. To do so, it uses system call accept. A call to accept blocks until a connection request arrives.

do {

new_sock = accept (s_sock, (struct sockaddr *) &client_adr, &client_len);

if (fork() == 0) { /* in CHILD process */

while ((len=read(new_sock, buf, size)){

/* process message */

} close (new_sock); exit(0);

} else close (new_sock); /* in parent */

} while (1); /* process messages forever */


On Demand

Collection

(server socket)




listen (s_sock, 5);

new_sock = accept (s_sock, (struct sockaddr *) &client_adr, &client_len);

When a request arrives, the system fills in client_adr with the address of the client that placed the request.

The system creates a new socket file descriptor, new_sock, that has its destination connected to the requesting client.


On Demand

Collection

(socket server)

Java applet

(socket client)

Active socket

(returned by accept)

When the server accepts a connection from a client, a new file descriptor is assigned.

Known port Number

Listening/

Passive socket


On Demand

Collection

(socket server)

Java applet

(socket client)

Known port

number

Second Client


Question: How can we have a single process handle all the client interfaces and at the same time do all PM count retrieval from the cells?

Why would we want to do this?


void ODinit()

{

/* Call the signal handler if we get one a SIGPOLL signal. This indicates that

** a socket needs servicing.

*/

(void) sigset (SIGPOLL, odsig_handler);

--- continued later ---

}

static void ODsig_handler (int sig)

{

switch (sig) {

case SIGPOLL:

sigpoll_flag = TRUE;

default: break;

}

}


main ( )

{ /* On Demand collection process */

ODinit(); /* initialize the socket *

while (1) {

while (ODwork_to_do()){

ODdo_work();

}

ODwait_for_work();

}

}


Void ODwait_for_work ( void ) /* Define how to react to Asynchronous Events - messages */

{ uxretval = UXWGETMSG ( &uxmbuf, UXFOREVER );

switch ( uxretval) {

case UXMSG: /* Process the UXMSG received from a cell - Network Element */

ODuxmsg(&uxmbuf );

break;

case UXEINTR: /* Got an interrupt.

Handled on return from this function */

break;

case UXENOPORT: /* Somehow the port we are doing

the UXWGETMSG on got dropped */

/* Restart the Unix port */

uxretval = UXCONNPORT(ITPT_OD);

break;

default: break;

}

}


int ODwork_to_do (void)

{

if (cl1timerflag || cl2timerflag || … )

return (TRUE);

if (rsp1timerflag || rsp2timerflag || … )

return (TRUE);

if (sigpoll_flag)

return (TRUE);

return (FALSE);

}


Void ODdo_work ( void ) {

if (sigpoll_flag) { /* Process client request */

sigpoll_flag = FALSE;

ODclient_req(); remember this one

}

if (cl1timerflag) { /* Process periodic request to Network elements (cells, ..)*/

ODmsgdest (ODCL1TIMERTAG);

odcl1timerflag = FALSE;

}

if (rsp1timerflag) { /* Timed out waiting for response from Network elements (cells, ..)*/

ODerr(ODRSP1TIMERTAG);

ODRSP1TIMERTAG = FALSE;

}

if (audit_timer_flag) { /* Time for a socket audit */

ODsock_audit();

}

}


ODinit continued

Void ODinit ()

{

/* Initialize the server structure */

server.sin_addr.s_addr = INADDR_ANY;

server.sin_family = AF_INET;

/* Get the service port associated with this process

usually in some /etc/ config file such as /etc/services */

servent_p = getservbyname ( PMSERVICE, “tcp” ); /* NOTE: API is not IPv6 compliant!*/

server.sin_port = servent_p->s_port;

/* Now create the connection socket */

ODsock = socket(AF_INET, SOCK_STREAM, IPPROTO_TCP);

/* Allow local address reuse */

on = 1;

setsockopt (ODsock, SOL_SOCKET, SO_REUSEADDR, (char *)&on, sizeof(on));


ODinit continued

Void ODinit ()

{

/* Bind the socket to the local address specified on the server.

This establishes a local address for the socket. */

bind (ODsock, (struct sockaddr *) (&server), sizeof(server));

/* Set up the server to listen for connection requests from the client.

This simply marks the socket as ready to receive a connection.

It does not wait for the client to connect. */

(void) listen (ODsock, 10); /* allow 10 connections pending */


Question: How do we know how many client to allow?


X X X

X X X X X

select() says these file descriptors from the ODreadmask have work pending

ODreadmask Set by FD_SET

FD_ISSET – tells if a particular file descriptor returned by select() has pending work


ODinit continued

Void ODinit ()

{

/* Include the socket descriptor in the list of sockets to poll. Set up

the descriptor so that we get a SIGPOLL when there’s something

to read from the socket. */

FD_ZERO (&ODreadmask);

FD_SET (ODsock, &ODreadmask);

if ( (rc -= ioctl (ODsock, I_SETSIG, S_INPUT )) == -1 ) {

error

}

/* At this point, if we get any input requests on the connection server

socket, we’ll get a SIGPOLL, call our sigpoll handler, and set a flag

indicating that we have input on one of our sockets.

*/


Void ODclient_req ( void )

{

/* ODreadmask is a global file descriptor set. Initially it includes only the

bound (listening) socket file descriptor. New file descriptors are added to the

ODreadmask after accepting a connection request.

Since select() modifies the read mask, we’ll have to make a copy. */

memcpy (&readfdset, &ODreadmask, sizeof(readfdset));

/* select() is used to determine if there are outstanding requests at

any of the socket fds */

numrequests = select (nfds, &readfdset, (fd_set *)NULL,

(fd_set *)NULL, &timeout);


ODclient continued

/* Loop through all the client socket file descriptors to determine

the next socket with an outstanding request. */

for (i = 0; i < nfds; i++) {

if (!FD_ISSET (socknum, &readfdset) {

/* nothing for this fd */

return;

}

/* If the request is from the listening socket, do an

accept and add the new socket file descriptor to the global readmask.

otherwise we have work for a client already connected. */

if (socknum == ODsock) {

ODconnect_client(); /* New client */

} else { ODapi (socknum); }

}


Void ODconnect_client ( void )

{

/* accept() will extract the first connection on the queue of pending

connections, create a new socket with the properties of ODsock,

and allocate a new file descriptor (clientsock) for the socket. */

clientsock = accept (ODsock, (struct sockaddr *)0, &i );

/* set up this socket so that it generates a SIGPOLL when it has input */

rc = ioctl (clientsock, I_SETSIG, S_INPUT);

/* Add the socket to the global read mask */

FD_SET ( clientsock, &ODreadmask );

/* We have a valid client. Send a challenge message to the client. */

ODsendchallenge ( clientsock, challenge_type );

}


Void ODsock_audit ()

{

/* Loop through connection table - list of all active socket file desriptors*/

/* If fd is open, but no activity for two intervals, close the socket */

ODclose_sock ( ODconnect[i], sockfd );

/* Reset the audit timer */

UXTIMER(intvl, AUDIT_TAG, ODtimer_exp);

}


Java applet – client side

public void run() {

// open a socket to the server

sock.open(hostName, portNo);

}

class PModSW {

public int open(String hostName, int portNo)

{

sock = new Socket(hostName, portNo)

// getInputStream is the input stream for reading bytes from the

// DataInputStream

sockIn = new DataInputStream(sock.getInputStream());

//getOutputStream writes to the DataOutputstream

sockOut = new DataOutputStream (sock.getOutputStream());

}


Java applet – client side continued

public String getLine()

{

// read response message from the OMP,

// waits until a message is received

line = sockIn.readLine();

}

public int sendReq (byte[] msg, int offset, int length)

{

// send a message to the server on the OMP

sockOut.write (msg, offset, length)

}

} // end of PModSW class

Additional Socket APIs- will be discussed in a future lecture


Socket Options

Various attributes that are used to determine the behavior of sockets.

Setting options tells the OS/Protocol Stack the behavior we want.

Support for generic options (apply to all sockets) and protocol specific options.


Setting and Getting option values

getsockopt() gets the current value of a socket option.

setsockopt() is used to set the value of a socket option.

#include <sys/socket.h>


int getsockopt( int sockfd,

int level,

int optname,

void *opval,

socklen_t *optlen);

level specifies whether the option is a general option or a protocol specific option (what level of code should interpret the option).

getsockopt()


int setsockopt( int sockfd,

int level,

int optname,

const void *opval,

socklen_t optlen);

setsockopt()


SO_REUSEADDR

Boolean option: enables binding to an address (port) that is already in use.

Used by servers that are transient - allows binding a passive socket to a port currently in use (with active sockets) by other processes.

Can be used to establish separate servers for the same service on different interfaces (or different IP addresses on the same interface).

Virtual Web Servers can work this way.


Signal Driven I/O (R. Stevens, text, chap.22 )

A signal handler can be installed to instruct the kernel to generate a SIGIO or SIGPOLL signal whenever something happens to a socket descriptor.

The signal handler must determine what conditions caused the signal and take appropriate action.

See also Stevens, sect. 7.10 – fcntl (SVR4) and ioctl(BSDx) and fcntl (POSIX)


Signal Driven UDP

SIGIO occurs whenever:– an incoming datagram arrives.

– an asynchronous error occurs.

» such as unreachable or invalid address

Could allow process to handle other tasks and still watch for incoming UDP messages.

Example of signal-driven UDP example described in the book ( R. Stevens, text, chap.22):– NTP Network Time Protocol.

– Used to record timestamp of arrival of UDP datagram.


Signal Driven TCP

SIGIO occurs whenever:– an incoming connection has completed.

– Disconnect request initiated.

– Disconnect request completed.

– Half a connection shutdown.

– Data has arrived.

– Data has been sent (indicating there is buffer space)

– Asynchronous error

tdc561 network programming week 4: client/server design approaches case study using sockets – a...

Documents

network programming

partial slide

clientserver programming

duplicated slide

small requests

child process

lucent technologies

future lecture slide