CSS432: End-to-End Protocols 1 CSS432 End-to-End Protocols Textbook Ch5.1 – 5.2 Professor: Munehiro Fukuda

2 IP Protocols Underlying best-effort network  drop messages  re-orders messages  delivers duplicate copies of a given message  limits messages to some finite size  delivers messages after an arbitrarily long delay M5, M4, M3, M2, M1M5, M2, M1 M4, M3 M4 M3 M2, M1, M4M5 Request a retransmission of M3 and M5 M5, and M3 M5, M5, M3 CSS432: End-to-End Protocols

3 Simple Demultiplexor (UDP) Unreliable and unordered datagram service Adds multiplexing No flow control Endpoints identified by ports  servers have well-known ports  see /etc/services on Unix Header format Optional checksum  psuedo header + UDP header + data SrcPortDstPort ChecksumLength Data CSS432: End-to-End Protocols

4 UDP: Simple Demultiplexer int sd; sd = socket(AF_INET, SOCK_DGRAM, 0); struct sockaddr_in server; server.sin_family =AF_INET; server.sin_addr.s_addr = htonl( INADDR_ANY) server.sin_port = htons( ); bind( sd, (struct sockaddr *)&server, sizeof( server ) ); recv( sd, buf, sizeof( buf ), 0 ); struct sockaddr_in server; struct hostent *hp, *gethostbyname( ); Server.sin_family = AF_INET; hp = gethostbyname( ); bcopy( hp->h_addr, &( server.sin_addr.s_addr ), sizeof( hp->h_length) ); server.sin_port = htons( ); sendto( sd, buf, sizeof( buf ), 0, (struct sockaddr *)&server, sizeof( server ) ); socket() sendto() socket() bind() recv() Port:12345 socket() bind() recv() Port: UDP Packets demultiplexed M5, M4, M3, M2, M1 CSS432: End-to-End Protocols

5 End-to-End Protocols Common end-to-end services  guarantee message delivery  deliver messages in FIFO order  deliver at most one copy of each message  support arbitrarily large messages  support synchronization  allow the receiver to flow control the sender  support multiple application processes on each host P1P2P3P4 Network M5, M4, M3, M2, M1 m5, m4, m3, m2, m1 CSS432: End-to-End Protocols

6 TCP Overview Connection-oriented Byte-stream  app writes bytes  TCP sends segments  app reads bytes Application process Write bytes TCP Send buffer Segment Transmit segments Application process Read bytes TCP Receive buffer … …… Full duplex Flow control: keep sender from overrunning receiver Congestion control: keep sender from overrunning network CSS432: End-to-End Protocols

7 Sockets (Code Example) int sd, newSd; sd = socket(AF_INET, SOCK_STREAM, 0); sockaddr_in server; bzero( (char *)&server, sizeof( server ) ); server.sin_family =AF_INET; server.sin_addr.s_addr = htonl( INADDR_ANY ) server.sin_port = htons( ); bind( sd, (sockaddr *)&server, sizeof( server ) ); struct hostent *host = gethostbyname( arg[0] ); sockaddr_in server; bzero( (char *)&server, sizeof( server ) ); server.sin_family = AF_INET; server.s_addr = inet_addr( inet_ntoa( *(struct in_addr*)*host->h_addr_list ) ); server.sin_port = htons( ); connect( sd, (sockaddr *)&server, sizeof( server ) ); sockaddr_in client; socklen_t len=sizeof(client); while( true ) { newSd = accept(sd, (sockaddr *)&client, &len); write( sd, buf1, sizeof( buf ) ); write( sd, buf2, sizeof( buf ) ); if ( fork( ) == 0 ) { close( sd ); read( newSd, buf1, sizeof( buf1 ) ); read( newSd, buf2, sizeof( buf2 ) ); } close( newSd ); } read() socket() connect() write() socket() connect() write() listen( sd, 5 ); socket() bind() listen() accept() buf2, buf1 close( newsd); exit( 0 ); int sd = socket(AF_INET, SOCK_STREAM, 0); CSS432: End-to-End Protocols

8 Data Link Versus Transport Potentially connects many different hosts  need explicit connection establishment and termination Potentially different RTT (Round Trip Time)  need adaptive timeout mechanism Potentially long delay in network  need to be prepared for arrival of very old packets  need to set MSL (Maximum Segment Lifetime) Potentially different capacity at destination  need to accommodate different node capacity Potentially different network capacity  need to be prepared for network congestion CSS432: End-to-End Protocols

9 Segment Format Each connection identified with 4-tuple:  (SrcPort, SrcIPAddr, DsrPort, DstIPAddr) Sliding window + flow control  Acknowledgment, SequenceNum, AdvertisedWinow Flags  SYN, FIN, RESET, PUSH, URG, ACK SYN: Establishing a conneciton FIN: terminating a connection RESET: Confused and Terminating PUSH: Section URG: Sending urgent data ACK: Validating acknowledgment field SequenceNum is incremented in all cases other than ACK. Sender Data (SequenceNum) Acknowledgment + AdvertisedWindow Receiver CSS432: End-to-End Protocols

10 Connection Establishment and Termination Active participant (client) Passive participant (server) SYN, SequenceNum = x SYN + ACK, SequenceNum = y, ACK, Acknowledgment = y + 1 Acknowledgment = x + 1 Tree-Way Handshake  Client Initiate a connection to a server with seq=x Set a timer and retransmit the request upon an expiration  Server Acknowledge the client request with ack=++x Initiate a reverse connection with seq=y Set a timer and retransmit the request upon an expiration  Client Acknowledge the server request with ack=++y X and y are chosen at random  Protect against two incarnations of the same connection using the same sequence number. CSS432: End-to-End Protocols

11 State Transition Diagram CLOSED LISTEN SYN_RCVDSYN_SENT ESTABLISHED CLOSE_WAIT LAST_ACKCLOSING TIME_WAIT FIN_WAIT_2 FIN_WAIT_1 Passive openClose Send/SYN SYN/SYN + ACK SYN + ACK/ACK SYN/SYN + ACK ACK Close/FIN FIN/ACKClose/FIN FIN/ACK ACK + FIN/ACK Timeout after two segment lifetimes (2 * MSL) FIN/ACK ACK Close/FIN Close CLOSED Active open/SYN Open  Active open A client connect( )  Passive open A server listen( ) Can change active Close  Active close A client or a server First close( ) Both side can be active  Passive close A client or server close( ) in response to the first close( ) CSS432: End-to-End Protocols

12 State Transition Diagram In what condition can the state transit from FIN_WAIT_1 to TIME_WAIT? What is the purpose of the TIME_WAIT state? CSS432: End-to-End Protocols

13 Timing Chart ClientServer ( connect( ) ) SYN_SENT SYN seq=x ACK=y + 1 SYN_RCVD ESTABLISHED LISTEN ( listen( ) ) SYN seq=y, ACK=x + 1 ESTABLISHED ( write( ) ) ( read( ) ) seq=x+1 ACK=y + 1 ACK x + 2 ( close( ) ) FIN_WAIT_1 FIN seq=x+2 ACK=y + 1 CLOSE_WAIT LAST_ACK( close( ) ) ACK x + 3 FIN seq = y + 1 FIN_WAIT_2 TIME_WAIT ACK=y + 2 Establishment Termination Data Transfer Peek such a flow with tcpdump in assignment 3. CSS432: End-to-End Protocols

14 Sliding Window Revisited Sending side  LastByteAcked ≤ LastByteSent  LastByteSent ≤ LastByteWritten  buffer bytes between [LastByteAcked, LastByteWritten] Sending application LastByteWritten TCP LastByteSentLastByteAcked Receiving application LastByteRead TCP LastByteRcvdNextByteExpected Receiving side  LastByteRead < NextByteExpected  NextByteExpected ≤ LastByteRcvd+1  buffer bytes between [ NextByteRead, LastByteRcvd] CSS432: End-to-End Protocols

15 Flow Control Sending application LastByteWritten TCP LastByteSent LastByteAcked Receiving application LastByteRead TCP LastByteRcvdNextByteExpected LastByteRcvd – LastByteRead ≤ MaxRcvbuffer AdvertisedWindow = MaxRcvBuffer – (NextByteExpected – NextByteRead) LastByteSent – LastByteAcked ≤ AdvertisedWindow EffectiveWindow = AdvertisedWindow – (LastByteSent – LastByteAcked) LastByteWritten – LastByteAcked ≤ MaxsendBuffer block sender if (LastByteWritten - LastByteAcked) + y > MaxSenderBuffer y Send ACK with an advertise window in response to arriving data segments as long as all the preceding bytes have also arrived and until the advertised window reaches 0. (ACK returned at the first time when it reaches 0) CSS432: End-to-End Protocols

16 Flow Control with A Slower Receiver Sending application LastByteWritten TCP LastByteSent LastByteAcked Receiving application LastByteRead TCP LastByteRcvd NextByteExpected LastByteRcvd – LastByteRead ≤ MaxRcvbuffer AdvertisedWindow = MaxRcvBuffer – (NextByteExpected – LastByteRead) LastByteSent – LastByteAcked ≤ AdvertisedWindow EffectiveWindow = AdvertisedWindow – (LastByteSent – LastByteAcked) LastByteWritten – LastByteAcked ≤ MaxsendBuffer block sender since (LastByteWritten - LastByteAcked) + y > MaxSenderBuffer < 0 yy No more send, no more ack, thus it stays In the same value Read slow. CSS432: End-to-End Protocols

17 The sender won’t send any more data. The receiver won’t initiate to send any advertised window. Then, how can the sender find out when the receiver can receive more data? Flow Control CSS432: End-to-End Protocols

18 32-bit SequenceNum  MSL (Maximum Segment Lifetime) = 120sec.  SequenceNum should not be wrapped around within 120 seconds. BandwidthTime Until Wrap Around T1 (1.5 Mbps)6.4 hours Ethernet (10 Mbps)57 minutes T3 (45 Mbps)13 minutes FDDI (100 Mbps)6 minutes STS-3 (155 Mbps)4 minutes STS-12 (622 Mbps)55 seconds STS-24 (1.2 Gbps)28 seconds Protection Against Wrap Around CSS432: End-to-End Protocols

19 Keeping the Pipe Full 16-bit AdvertisedWindow = 64KB Assuming that RTT = 100msec, the advertised window’s capacity is at least RTT x Network bandwidth BandwidthRTT(100msec) x Bandwidth Product T1 (1.5 Mbps)18KB Ethernet (10 Mbps)122KB T3 (45 Mbps)549KB FDDI (100 Mbps)1.2MB STS-3 (155 Mbps)1.8MB STS-12 (622 Mbps)7.4MB STS-24 (1.2 Gbps)14.8MB CSS432: End-to-End Protocols

20 Segment Transmission A segment is transmitted out: 1. When a segment to send reaches Maximum segment size (MMS) = Maximum Transfer Unit (MTU) 2. When a TCP receives a push operation that flushes the unsent data (Peek with tcpdump in programming assignment 3) 3. When a timer expires A trigger by a timer may cause the silly window syndrome. CSS432: End-to-End Protocols

21 Silly Window Syndrome If a sender aggressively takes advantage of any available window,  The receiver empties every window regardless of its size and thus small windows will never disappear. The problem occurs only when either the sender transmits a small segment or the receiver opens the window a small amount The receiver can delay ACKs to make a larger window  How long does it wait? The sender should make a decision  Nagle’s Algorithm (Programming assignment 3) MMS 1 MMS 2 Ad Window small SenderReceiver CSS432: End-to-End Protocols

22 Nagle’s Algorithm Ack works as a timer to fire a new segment transmission. The algorithm may not respond to interactive or real-time applications  TCP_NODELAY Option: Transmit data as soon as possible setsockopt(sockfd, SOL_TCP, TCP_NODELAY, &intFlag, sizeof(intFlag)) When the application produces data to send { if both the available data and the window ≥ MSS send a full segment // no problem, send it right now else if there is unAcked data in transit // a full empty segment returns soon buffer the new data until an ACK arrives else // the receiver won’t reply. Can’t avoid the silly window syndrome send all the new data now } CSS432: End-to-End Protocols

23 How to Estimate RTT Timeout is necessary  To retransmit a lost packet  To detect congestions and invoke AIMD. RTT estimation algorithms  Original Algorithm  Karn/Partridge Algorithm  Jacobson/Karels Algorithm CSS432: End-to-End Protocols

24 Original Algorithm Measure SampleRTT for each segment/ ACK pair Compute weighted average of RTT  EstRTT =  x EstRTT +  x SampleRTT  where  +  = 1   between 0.8 and 0.9   between 0.1 and 0.2 Set timeout based on EstRTT  TimeOut = 2 x EstRTT  Why double? EstRTT cannot respond to deviated SampleRTT quickly. CSS432: End-to-End Protocols

25 Karn/Partridge Algorithm Do not sample RTT when retransmitting  Can’t figure out which transmission the latest ACK corresponds to. Double timeout after each retransmission  Congestion causes this retransmission.  Modestly retransmit segments. SenderReceiver Original transmission ACK SampleR TT Retransmission SenderReceiver Original transmission ACK SampleR TT Retransmission CSS432: End-to-End Protocols

26 Jacobson/ Karels Algorithm Original Algorithm  EstRTT =  x EstRTT +  x SampleRTT 0.8 and and 0.2  TimeOut = EstRTT * 2 New Algorithm that takes into a consideration if the variation among smapleRTTs are large.  EstRTT = EstRTT +  (SampleRTT – EstRTT) Diff  Dev = Dev +  (|SampleRTT – EstRTT| – Dev) Diff  TimeOut =  x EstRTT +  x Dev 1 4 CSS432: End-to-End Protocols

27 Jacobson/ Karels Algorithm SampleRTT –= (EstimatedRTT >> 3) EstimatedRTT += SampleRTT; If (SampleRTT < 0) SampleRTT = – SampelRTT; SampleRTT –= (Deviation >> 3); Deviation += SampleRTT; TimeOut = (EstimatedRTT >> 3) + (Deviation >> 1) EstimatedRTT = 8 EstRTT Deviation = 8 Dev SampleRTT = (SampleRTT – 8 EstRTT) / 8 = SampleRTT – EstRTT) diff 8 EstRTT = 8EstRTT + SampleRTT = 8EstRTT + (SampleRTT – EstRTT) EstRTT = EstRTT + 1/8 (SampleRTT – EstRTT) diff 8 Dev = 8 Dev + |SampleRTT – EsRTT| - Dev Dev = Dev + 1/8( |SampleRTT – EstRTT| - Dev ) TimeOut = 8 EstRTT / Dev / 2 TimeOut = EstRTT + 4 Dev |SampleRTT – EstRTT| – 8Dev/8 |diff| Notes  Algorithm does not work accurately with a large granularity of clock (500ms on Unix)  Accurate timeout mechanism important to congestion control CSS432: End-to-End Protocols

28 TCP Extensions RTT Measurement  Store a 32-bit timestamp in outgoing segments’ header option  Receive an ack with the original timestamp  Sample RTT by subtracting the timestamp from the current timer Resolving the quick wrap-around of sequence number  The 32-bit timestamp and the 32-bit sequence number gives a 64-bit sequence space Extending an advertised window  Scale up the advertised window How many bytes to send → How many 16-byte units to send CSS432: End-to-End Protocols

29 Reviews  UDP  TCP: three-way handshake and state transition  Sliding window and flow control  Segment transmission: silly window syndrome and Nagle’s algorithm  Adaptive retransmission: original, Karn/Partridge, and Jacobson/Karels Exercises in Chapter 5  Ex. 5, 14, 22, and 39 (TCP state transition)  Ex. 9(a) (Sliding window)  Ex. 20 (Nagle’s algorithm) CSS432: End-to-End Protocols