Connections. Topics. Focus. Presentation Session. Application. Data Link. Transport. Physical. Network

Transcription

1 Connections Focus How do we connect processes? This is the transport layer Topics Naming processes Connection setup / teardown Flow control Application Presentation Session Transport Network Data Link Physical djw // CSE 461, Fall 2009 Lconn.1

2 The Transport Layer Builds on the services of the Network layer TCP/IP Communication between processes running on hosts Naming/Addressing Stronger guarantees of message delivery make sense This is the first layer that is talking end-to-end djw // CSE 461, Fall 2009 Lconn.2

3 Internet Transport Protocols UDP Datagram abstraction between processes With error detection SrcPort DstPort Length Checksum Data TCP Bytestream abstraction between processes With reliability Plus congestion control (later!) djw // CSE 461, Fall 2009 Lconn.3

4 Comparison of TCP/UDP/IP properties TCP Connection-oriented Reliable byte-stream In-order delivery Single delivery Arbitrarily length Synchronization Flow control Congestion control UDP Datagram oriented Lost packets Reordered packets Duplicate packets Limited size packets IP Datagram oriented Lost packets Reordered packets Duplicate packets Limited size packets djw // CSE 461, Fall 2009 Lconn.4

5 Naming Processes/Services Process here is an abstract term for your Web browser (HTTP), servers (SMTP), hostname translation (DNS), RealAudio player (RTSP), etc. How do we identify for remote communication? Process id or memory address are OS-specific and transient So TCP and UDP use Ports 16-bit integers representing mailboxes that processes rent Identify process uniquely as (IP address, protocol, port) djw // CSE 461, Fall 2009 Lconn.5

6 Processes as Endpoints djw // CSE 461, Fall 2009 Lconn.6

7 Picking Port Numbers We still have the problem of allocating port numbers What port should a Web server use on host X? To what port should you send to contact that Web server? Servers typically bind to well-known port numbers e.g., HTTP 80, SMTP 25, DNS 53, look in /etc/services Ports below 1024 reserved for well-known services Clients use OS-assigned temporary (ephemeral) ports Above 1024, recycled by OS when client finished djw // CSE 461, Fall 2009 Lconn.7

8 User Datagram Protocol (UDP) Provides message delivery between processes Source port filled in by OS as message is sent Destination port identifies UDP delivery queue at endpoint SrcPort DstPort Checksum Length Data djw // CSE 461, Fall 2009 Lconn.8

9 UDP Delivery Ports Application process Application process Application process Kernel boundary Message Queues DeMux Packets arrive djw // CSE 461, Fall 2009 Lconn.9

10 UDP Checksum UDP includes optional protection against errors Checksum intended as an end-to-end check on delivery So it covers data, UDP header, and IP pseudoheader SrcPort DstPort Checksum Length Data djw // CSE 461, Fall 2009 Lconn.10

11 Transmission Control Protocol (TCP) Reliable bi-directional bytestream between processes Message boundaries are not preserved Connections Conversation between endpoints with beginning and end Flow control (later) Prevents sender from over-running receiver buffers Congestion control (later) Prevents sender from over-running network buffers djw // CSE 461, Fall 2009 Lconn.11

12 TCP Delivery Application process Application process Write bytes Read bytes TCP Send buffer TCP Receive buffer Transmit segments Segment Segment Segment djw // CSE 461, Fall 2009 Lconn.12

13 TCP Header Format Ports plus IP addresses identify a connection SrcPort DstPort SequenceNum Acknowledgment HdrLen 0 Flags AdvertisedWindow Checksum UrgPtr Options (variable) Data djw // CSE 461, Fall 2009 Lconn.13

14 TCP Header Format Sequence, Ack numbers used for the sliding window Congestion control works by controlling the window size SrcPort DstPort SequenceNum Acknowledgment HdrLen 0 Flags AdvertisedWindow Checksum UrgPtr Options (variable) Data djw // CSE 461, Fall 2009 Lconn.14

15 TCP Header Format Flags may be URG, ACK, PSH, RST, SYN, FIN SrcPort DstPort SequenceNum Acknowledgment HdrLen 0 Flags AdvertisedWindow Checksum UrgPtr Options (variable) Data djw // CSE 461, Fall 2009 Lconn.15

16 TCP Header Format Advertised window is used for flow control SrcPort DstPort SequenceNum Acknowledgment HdrLen 0 Flags AdvertisedWindow Checksum UrgPtr Options (variable) Data djw // CSE 461, Fall 2009 Lconn.16

17 Other TCP Header Fields Header length allows for variable length TCP header with options for extensions such as timestamps, selective acknowledgements, etc. Checksum is analogous to that of UDP Urgent pointer/data not used in practice Very few bits not assigned djw // CSE 461, Fall 2009 Lconn.17

18 Connection Establishment Both sender and receiver must be ready before we start to transfer the data Sender and receiver need to agree on a set of parameters e.g., the Maximum Segment Size (MSS) This is signaling It sets up state at the endpoints Compare to dialing in the telephone network In TCP a Three-Way Handshake is used djw // CSE 461, Fall 2009 Lconn.18

19 Three-Way Handshake Opens both directions for transfer Active participant (client) Passive participant (server) SYN, SequenceNum = x SYN + ACK, SequenceNum = y, Acknowledgment = x + 1 ACK, Acknowledgment = y + 1 +data djw // CSE 461, Fall 2009 Lconn.19

20 Some Comments We could abbreviate this setup, but it was chosen to be robust, especially against delayed duplicates Three-way handshake from Tomlinson 1975 Choice of changing initial sequence numbers (ISNs) minimizes the chance of hosts that crash getting confused by a previous incarnation of a connection But with random ISN it actually proves that two hosts can communicate Weak form of authentication djw // CSE 461, Fall 2009 Lconn.20

21 TCP State Transitions CLOSED Active open/syn Passive open Close Close LISTEN SYN/SYN + ACK Send/ SYN SYN/SYN + ACK SYN_RCVD SYN_SENT ACK SYN + ACK/ACK Close /FIN ESTABLISHED FIN_WAIT_1 Close /FIN FIN/ACK FIN/ACK CLOSE_WAIT FIN_WAIT_2 CLOSING LAST_ACK ACK Timeout after two ACK segment lifetimes FIN/ACK TIME_WAIT CLOSED djw // CSE 461, Fall 2009 Lconn.21 ACK + FIN/ACK ACK Close /FIN

22 Again, with States Active participant (client) SYN_SENT Passive participant (server) LISTEN SYN, SequenceNum = x SYN + ACK, SequenceNum = y, Acknowledgment = x + 1 ESTABLISHED SYN_RCVD ACK, Acknowledgment = y + 1 +data ESTABLISHED djw // CSE 461, Fall 2009 Lconn.22

23 Connection Teardown Orderly release by sender and receiver when done Delivers all pending data and hangs up Cleans up state in sender and receiver TCP provides a symmetric close both sides shutdown independently djw // CSE 461, Fall 2009 Lconn.23

24 TCP Connection Teardown Web server Web browser FIN FIN_WAIT_1 CLOSE_WAIT LAST_ACK CLOSED ACK FIN ACK FIN_WAIT_2 TIME_WAIT CLOSED djw // CSE 461, Fall 2009 Lconn.24

25 The TIME_WAIT State We wait 2MSL (two times the maximum segment lifetime of 60 seconds) before completing the close Why? ACK might have been lost and so FIN will be resent Could interfere with a subsequent connection djw // CSE 461, Fall 2009 Lconn.25

26 Berkeley Sockets interface Networking protocols implemented in OS OS must expose a programming API to applications most OSs use the socket interface originally provided by BSD 4.1c in ~1982. Principle abstraction is a socket a point at which an application attaches to the network defines operations for creating connections, attaching to network, sending and receiving data, closing connections djw // CSE 461, Fall 2009 Lconn.26

27 TCP (connection-oriented) Server Socket() Block until connect Process request Bind() Listen() Accept() Recv() Send() Connection Establishmt. Data (request) Data (reply) Client Socket() Connect() Send() Recv() djw // CSE 461, Fall 2009 Lconn.27

28 UDP (connectionless) Server Block until Data from client Socket() Bind() Recvfrom() Data (request) Client Socket() Bind() Sendto() Process request Sendto() Data (reply) Recvfrom() djw // CSE 461, Fall 2009 Lconn.28

29 Diversion: TCP SYN cookies Solution: SYN cookies Server keeps no state in response to SYN; instead makes client store state Server picks return seq # y = that encrypts x Gets +1 from sender; unpacks to yield x Can data arrive before ACK? Client Server SYN, SequenceNum = x SYN + ACK,, x + 1 ACK, + 1 djw // CSE 461, Fall 2009 Lconn.29

30 Diversion: HTTP on TCP How do we reduce the # of messages? SYN SYN+ACK Delayed ack: wait for 200ms for reply or another pkt arrival ACK http get ACK TCP RST from web server http data ACK FIN ACK FIN ACK djw // CSE 461, Fall 2009 Lconn.30