Communication Networks/TCP and UDP Protocols/TCP
After going through the various layers of the Model, it’s time to have a look at the TCP protocol and to study its functionality. This section will help the reader to get to know about the concepts and characteristics of the TCP, and then gradually dive into the details of TCP like connection establishment/closing, communication in TCP and why the TCP protocol is called a reliable as well as an adaptive protocol. This section will end with a comparison between UDP and TCP followed by a nice exercise which would encourage readers to solve more and more problems.
Before writing this section, the information has been studied from varied sources like TCP guide, RFC's, tanenbaum book and the class notes.
What is TCP?
In theory, a transport layer protocol could be a very simple software routine, but the TCP protocol cannot be called simple. Why use a transport layer which is as complex as TCP? The most important reason depends on IP's unreliability. In fact all the layers below TCP are unreliable and deliver the datagram hop-by-hop. The IP layer delivers the datagram hop-by-hop and does not guarantee delivery of a datagram; it is a connectionless system. IP simply handles the routing of datagrams; and if problems occur, IP discards the packet without a second thought, generating an error message back to the sender in the process. The task of ascertaining the status of the datagrams sent over a network and handling the resending of information if parts have been discarded falls to TCP.
Most users think of TCP and IP as a tightly knit pair, but TCP can be, and frequently is, used with other transport protocols.
For example, TCP or parts of it are used in the File Transfer Protocol (FTP) and the Simple Mail Transfer Protocol (SMTP), both of which do not use IP.
The Transmission Control Protocol provides a considerable number of services to the IP layer and the upper layers. Most importantly, it provides a connection-oriented protocol to the upper layers that enable an application to be sure that a datagram sent out over the network was received in its entirety. In this role, TCP acts as a message-validation protocol providing reliable communications. If a datagram is corrupted or lost, it is usually TCP (not the applications in the higher layers) that handles the retransmission.
TCP is not a piece of software. It is a communications protocol.
TCP manages the flow of datagrams from the higher layers, as well as incoming datagrams from the IP layer. It has to ensure that priorities and security are respected. TCP must be capable of handling the termination of an application above it that was expecting incoming datagrams, as well as failures in the lower layers. TCP also must maintain a state table of all data streams in and out of the TCP layer. The isolation of these services in a separate layer enables applications to be designed without regard to flow control or message reliability. Without the TCP layer, each application would have to implement the services themselves, which is a waste of resources.
TCP resides in the transport layer, positioned above IP but below the upper layers and their applications, as shown in the Figure below. TCP resides only on devices that actually process datagrams, ensuring that the datagram has gone from the source to target machines. It does not reside on a device that simply routes datagrams, so there is no TCP layer in a gateway. This makes sense, because on a gateway the datagram has no need to go higher in the layered model than the IP layer.
Figure 1: TCP providing reliable End-to-End communication
Because TCP is a connection-oriented protocol responsible for ensuring the transfer of a datagram from the source to destination machine (end-to-end communications), TCP must receive communications messages from the destination machine to acknowledge receipt of the datagram. The term virtual circuit is usually used to refer to the handshaking that goes on between the two end machines, most of which are simple acknowledgment messages (either confirmation of receipt or a failure code) and datagram sequence numbers. It is analogous to a telephone conversation; someone initiates it by ringing a number which is answered, a two-way conversation takes place, and finally someone ends the conversation. A socket pair identifies both ends of a connection, i.e. the virtual circuit. It may be recalled that the socket consists of the IP address and the port number to identify the location. The Servers use well-known port numbers (< 1K) for standardized services (Listen). Numbers over 1024 are available for users to use freely. Port numbers for some of the standard services are given below in the table below.
|69||TFTP||Trivial file transfer protocol|
|79||Finger||Lookup information about a user|
|80||HTTP||World Wide Web|
|110||POP-3||Remote e-mail access|
Byte stream or Message Stream?
Well, the message boundaries are not preserved end to end in the TCP. For example, if the sending process does four 512-byte writes to a TCP stream, these data may be delivered to the receiving process as four 512-byte chunks, two 1024-byte chunks, one 2048-byte chunk), or some other way. There is no way for the receiver to detect the unit(s) in which the data were written. A TCP entity accepts user data streams from local processes, breaks them up into pieces not exceeding 64 KB (in practice, often 1460 data bytes in order to fit in a single Ethernet frame with the IP and TCP headers), and sends each piece as a separate IP datagram. When datagrams containing TCP data arrive at a machine, they are given to the TCP entity, which reconstructs the original byte streams. For simplicity, we will sometimes use just TCP to mean the TCP transport entity (a piece of software) or the TCP protocol (a set of rules). From the context it will be clear which is meant. For example, in The user gives TCP the data, the TCP transport entity is clearly intended. The IP layer gives no guarantee that datagrams will be delivered properly, so it is up to TCP to time out and retransmit them as need be. Datagrams that do arrive may well do so in the wrong order; it is also up to TCP to reassemble them into messages in the proper sequence. In short, TCP must furnish the reliability that most users want and that IP does not provide.
Characteristics of TCP
TCP provides a communication channel between processes on each host system. The channel is reliable, full-duplex, and streaming. To achieve this functionality, the TCP drivers break up the session data stream into discrete segments, and attach a TCP header to each segment. An IP header is attached to this TCP packet, and the composite packet is then passed to the network for delivery. This TCP header has numerous fields that are used to support the intended TCP functionality. TCP has the following functional characteristics:
Unicast protocol : TCP is based on a unicast network model, and supports data exchange between precisely two parties. It does not support broadcast or multicast network models.
Connection state : Rather than impose a state within the network to support the connection, TCP uses synchronized state between the two endpoints. This synchronized state is set up as part of an initial connection process, so TCP can be regarded as a connection-oriented protocol. Much of the protocol design is intended to ensure that each local state transition is communicated to, and acknowledged by, the remote party.
Reliable : Reliability implies that the stream of octets passed to the TCP driver at one end of the connection will be transmitted across the network so that the stream is presented to the remote process as the same sequence of octets, in the same order as that generated by the sender. This implies that the protocol detects when segments of the data stream have been discarded by the network, reordered, duplicated, or corrupted. Where necessary, the sender will retransmit damaged segments so as to allow the receiver to reconstruct the original data stream. This implies that a TCP sender must maintain a local copy of all transmitted data until it receives an indication that the receiver has completed an accurate transfer of the data.
Full duplex : TCP is a full-duplex protocol; it allows both parties to send and receive data within the context of the single TCP connection.
Streaming : Although TCP uses a packet structure for network transmission, TCP is a true streaming protocol, and application-level network operations are not transparent. Some protocols explicitly encapsulate each application transaction; for every write , there must be a matching read . In this manner, the application-derived segmentation of the data stream into a logical record structure is preserved across the network. TCP does not preserve such an implicit structure imposed on the data stream, so that there is no pairing between write and read operations within the network protocol. For example, a TCP application may write three data blocks in sequence into the network connection, which may be collected by the remote reader in a single read operation. The size of the data blocks (segments) used in a TCP session is negotiated at the start of the session. The sender attempts to use the largest segment size it can for the data transfer, within the constraints of the maximum segment size of the receiver, the maximum segment size of the configured sender, and the maximum supportable non-fragmented packet size of the network path (path Maximum Transmission Unit [MTU]). The path MTU is refreshed periodically to adjust to any changes that may occur within the network while the TCP connection is active.
Rate adaptation : TCP is also a rate-adaptive protocol, in that the rate of data transfer is intended to adapt to the prevailing load conditions within the network and adapt to the processing capacity of the receiver. There is no predetermined TCP data-transfer rate; if the network and the receiver both have additional available capacity, a TCP sender will attempt to inject more data into the network to take up this available space. Conversely, if there is congestion, a TCP sender will reduce its sending rate to allow the network to recover. This adaptation function attempts to achieve the highest possible data-transfer rate without triggering consistent data loss.
TCP Header structure
TCP segments are sent as Internet datagrams. The Internet Protocol header carries several information fields, including the source and destination host addresses. A TCP header follows the Internet header, supplying information specific to the TCP protocol. This division allows for the existence of host level protocols other than TCP.
0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Source Port | Destination Port | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Sequence Number | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Acknowledgment Number | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Data | |U|A|P|R|S|F| | | Offset| Reserved |R|C|S|S|Y|I| Window | | | |G|K|H|T|N|N| | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Checksum | Urgent Pointer | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Options | Padding | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | data | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ TCP Header Format Note that one tick mark represents one bit position.
Source Port: 16 bits The source port number.
Destination Port: 16 bits The destination port number.
Sequence Number: 32 bit The sequence number of the first data octet in this segment (except when SYN is present). If SYN is present the sequence number is the initial sequence number (ISN) and the first data octet is ISN+1.
Acknowledgment Number: 32 bits If the ACK control bit is set this field contains the value of the next sequence number the sender of the segment is expecting to receive. Once a connection is established this is always sent.
Data Offset: 4 bits The number of 32 bit words in the TCP Header. This indicates where the data begins. The TCP header (even one including options) is an integral number of 32 bits long.
Reserved: 6 bits Reserved for future use. Must be zero.
Control Bits: 6 bits (from left to right):
URG: Urgent Pointer field significant
ACK: Acknowledgment field significant
PSH: Push Function
RST: Reset the connection
SYN: Synchronize sequence numbers
FIN: No more data from sender
Window: 16 bits The number of data octets beginning with the one indicated in the acknowledgment field which the sender of this segment is willing to accept.
Checksum: 16 bits The checksum field is the 16 bit one's complement of the one's complement sum of all 16 bit words in the header and text. If a segment contains an odd number of header and text octets to be checksummed, the last octet is padded on the right with zeros to form a 16 bit word for checksum purposes. The pad is not transmitted as part of the segment. While computing the checksum, the checksum field itself is replaced with zeros.
The checksum also covers a 96 bit pseudo header conceptually prefixed to the TCP header. This pseudo header contains the Source Address, the Destination Address, the Protocol, and TCP length. This gives the TCP protection against misrouted segments. This information is carried in the Internet Protocol and is transferred across the TCP/Network interface in the arguments or results of calls by the TCP on the IP.
The TCP Length is the TCP header length plus the data length in octets (this is not an explicitly transmitted quantity, but is computed), and it does not count the 12 octets of the pseudo header.
Urgent Pointer: 16 bits This field communicates the current value of the urgent pointer as a positive offset from the sequence number in this segment. The urgent pointer points to the sequence number of the octet following the urgent data. This field is only be interpreted in segments with the URG control bit set.
Options: variable Options may occupy space at the end of the TCP header and are a multiple of 8 bits in length. All options are included in the checksum. An option may begin on any octet boundary. There are two cases for the format of an option:
Case 1: A single octet of option-kind.
Case 2: An octet of option-kind, an octet of option-length, and the actual option-data octets. The option-length counts the two octets of option-kind and option-length as well as the option-data octets. Note that the list of options may be shorter than the data offset field might imply. The content of the header beyond the End-of-Option option must be header padding (i.e., zero).
A TCP must implement all options
Communication in TCP
Before TCP can be employed for any actually useful purpose—that is, sending data—a connection must be set up between the two devices that wish to communicate. This process, usually called connection establishment, involves an exchange of messages that transitions both devices from their initial connection state (CLOSED) to the normal operating state (ESTABLISHED).
Connection Establishment Functions
The connection establishment process actually accomplishes several things as it creates a connection suitable for data exchange:
Contact and Communication: The client and server make contact with each other and establish communication by sending each other messages. The server usually doesn’t even know what client it will be talking to before this point, so it discovers this during connection establishment.
Sequence Number Synchronization: Each device lets the other know what initial sequence number it wants to use for its first transmission.
Parameter Exchange: Certain parameters that control the operation of the TCP connection are exchanged by the two devices.
Control Messages Used for Connection Establishment: SYN and ACK
TCP uses control messages to manage the process of contact and communication. There aren't, however, any special TCP control message types; all TCP messages use the same segment format. A set of control flags in the TCP header indicates whether a segment is being used for control purposes or just to carry data. Following flags are altered while using control messages.
SYN: This bit indicates that the segment is being used to initialize a connection. SYN stands for synchronize, in reference to the sequence number synchronization I mentioned above.
ACK: This bit indicates that the device sending the segment is conveying an acknowledgment for a message it has received (such as a SYN).
Normal Connection Establishment: The "Three Way Handshake"
To establish a connection, each device must send a SYN and receive an ACK for it from the other device. Thus, conceptually, four control messages need to be passed between the devices. However, it's inefficient to send a SYN and an ACK in separate messages when one could communicate both simultaneously. Thus, in the normal sequence of events in connection establishment, one of the SYNs and one of the ACKs is sent together by setting both of the relevant bits (a message sometimes called a SYN+ACK). This makes a total of three messages, and for this reason the connection procedure is called a three-way handshake.
Key Concept: The normal process of establishing a connection between a TCP client and server involves three steps:
the client sends a SYN message; the server sends message that combines an ACK for the client’s SYN and contains the server’s SYN; and then the client sends an ACK for the server’s SYN. This is called the TCP three-way handshake.
A connection progresses through a series of states during its lifetime.
The states are: LISTEN, SYN-SENT, SYN-RECEIVED,ESTABLISHED, FIN-WAIT-1, FIN-WAIT-2, CLOSE-WAIT, CLOSING, LAST-ACK, TIME-WAIT, and the fictional state CLOSED. CLOSED is fictional because it represents the state when there is no TCB, and therefore, no connection. Briefly the meanings of the states are:
LISTEN - represents waiting for a connection request from any remote TCP and port.
SYN-SENT - represents waiting for a matching connection request after having sent a connection request.
SYN-RECEIVED - represents waiting for a confirming connection request acknowledgment after having both received and sent a connection request.
ESTABLISHED - represents an open connection, data received can be delivered to the user. The normal state for the data transfer phase of the connection.
FIN-WAIT-1 - represents waiting for a connection termination request from the remote TCP, or an acknowledgment of the connection termination request previously sent.
FIN-WAIT-2 - represents waiting for a connection termination request from the remote TCP.
CLOSE-WAIT - represents waiting for a connection termination request from the local user.
CLOSING - represents waiting for a connection termination request acknowledgment from the remote TCP.
LAST-ACK - represents waiting for an acknowledgment of the connection termination request previously sent to the remote TCP (which includes an acknowledgment of its connection termination request).
TIME-WAIT - represents waiting for enough time to pass to be sure the remote TCP received the acknowledgment of its connection termination request.
CLOSED - represents no connection state at all.
A TCP connection progresses from one state to another in response to events. The events are the user calls, OPEN, SEND, RECEIVE, CLOSE, ABORT, and STATUS; the incoming segments, particularly those containing the SYN, ACK, RST and FIN flags; and timeouts.
The state diagram in figure 6 illustrates only state changes, together with the causing events and resulting actions, but addresses neither error conditions nor actions which are not connected with state changes. In a later section, more detail is offered with respect to the reaction of the TCP to events.
Key Concept: If one device setting up a TCP connection sends a SYN and then receives a SYN from the other one before its SYN is acknowledged, the two devices perform a simultaneous open, which consists of the exchange of two independent SYN and ACK message sets. The end result is the same as the conventional three-way handshake, but the process of getting to the ESTABLISHED state is different. The possibility of collision normally occurs in Peer-2-Peer connection.
Buffer management When the Sender(assume client in our case) has a connection to establish, the packet comes to the Transmission Buffer. The packet should have some sequence number attached to it. This sender chooses the sequence number to minimize the risk of using the already used sequence number. The client sends the packet with that sequence number and data along with the packet length field. The server on receiving the packet sends ACK of the next expected sequence number. It also sends the SYN with it’s own sequence number.
The client on receiving both the messages ( SYN as well as ACK), sends ACK to the receiver with the next expected sequence number from the Receiver. Thus, the sequence number are established between the Client and Server. Now, they are ready for the data transfer. Even while sending the data, same concept of the sequence number is followed.
TCP transmission Policy
The window management in TCP is not directly tied to acknowledgements as it is in most data link protocols. For example, suppose the receiver has a 4096-byte buffer, as shown in Figure below. If the sender transmits a 2048-byte segment that is correctly received, the receiver will acknowledge the segment. However, since it now has only 2048 bytes of buffer space (until the application removes some data from the buffer), it will advertise a window of 2048 starting at the next byte expected.
Now the sender transmits another 2048 bytes, which are acknowledged, but the advertised window is 0. The sender must stop until the application process on the receiving host has removed some data from the buffer, at which time TCP can advertise a larger window.
When the window is 0, the sender may not normally send segments, with two exceptions. First, urgent data may be sent, for example, to allow the user to kill the process running on the remote machine. Second, the sender may send a 1-byte segment to make the receiver reannounce the next byte expected and window size. The TCP standard explicitly provides this option to prevent deadlock if a window announcement ever gets lost.
Senders are not required to transmit data as soon as they come in from the application. Neither are receivers required to send acknowledgements as soon as possible. When the first 2 KB of data came in, TCP, knowing that it had a 4-KB window available, would have been completely correct in just buffering the data until another 2 KB came in, to be able to transmit a segment with a 4-KB payload. This freedom can be exploited to improve performance.
Consider a telnet connection to an interactive editor that reacts on every keystroke. In the worst case, when a character arrives at the sending TCP entity, TCP creates a 21-byte TCP segment, which it gives to IP to send as a 41-byte IP datagram. At the receiving side, TCP immediately sends a 40-byte acknowledgment (20 bytes of TCP header and 20 bytes of IP header). Later, when the editor has read the byte, TCP sends a window update, moving the window 1 byte to the right. This packet is also 40 bytes. Finally, when the editor has processed the character, it echoes the character as a 41-byte packet. In all, 162 bytes of bandwidth are used and four segments are sent for each character typed. When bandwidth is scarce, this method of doing business is not desirable.
One approach that many TCP implementations use to optimize this situation is to delay acknowledgments and window updates for 500 msec in the hope of acquiring some data on which to hitch a free ride. Assuming the editor echoes within 500 msec, only one 41-byte packet now need be sent back to the remote user, cutting the packet count and bandwidth usage in half. Although this rule reduces the load placed on the network by the receiver, the sender is still operating inefficiently by sending 41-byte packets containing 1 byte of data. A way to reduce this usage is known as Nagle's algorithm (Nagle, 1984). What Nagle suggested is simple: when data come into the sender one byte at a time, just send the first byte and buffer all the rest until the outstanding byte is acknowledged. Then send all the buffered characters in one TCP segment and start buffering again until they are all acknowledged. If the user is typing quickly and the network is slow, a substantial number of characters may go in each segment, greatly reducing the bandwidth used. The algorithm additionally allows a new packet to be sent if enough data have trickled in to fill half the window or a maximum segment.
Nagle's algorithm is widely used by TCP implementations, but there are times when it is better to disable it. In particular, when an X Windows application is being run over the Internet, mouse movements have to be sent to the remote computer. (The X Window system is the windowing system used on most UNIX systems.) Gathering them up to send in bursts makes the mouse cursor move erratically, which makes for unhappy users.
Another problem that can degrade TCP performance is the silly window syndrome. This problem occurs when data are passed to the sending TCP entity in large blocks, but an interactive application on the receiving side reads data 1 byte at a time. To see the problem, look at the figure below. Initially, the TCP buffer on the receiving side is full and the sender knows this (i.e., has a window of size 0). Then the interactive application reads one character from the TCP stream. This action makes the receiving TCP happy, so it sends a window update to the sender saying that it is all right to send 1 byte. The sender obliges and sends 1 byte. The buffer is now full, so the receiver acknowledges the 1-byte segment but sets the window to 0. This behavior can go on forever.
Clark's solution is to prevent the receiver from sending a window update for 1 byte. Instead it is forced to wait until it has a decent amount of space available and advertise that instead. Specifically, the receiver should not send a window update until it can handle the maximum segment size it advertised when the connection was established or until its buffer is half empty, whichever is smaller.
Furthermore, the sender can also help by not sending tiny segments. Instead, it should try to wait until it has accumulated enough space in the window to send a full segment or at least one containing half of the receiver's buffer size (which it must estimate from the pattern of window updates it has received in the past).
Nagle's algorithm and Clark's solution to the silly window syndrome are complementary. Nagle was trying to solve the problem caused by the sending application delivering data to TCP a byte at a time. Clark was trying to solve the problem of the receiving application sucking the data up from TCP a byte at a time. Both solutions are valid and can work together. The goal is for the sender not to send small segments and the receiver not to ask for them.
The receiving TCP can go further in improving performance than just doing window updates in large units. Like the sending TCP, it can also buffer data, so it can block a READ request from the application until it has a large chunk of data to provide. Doing this reduces the number of calls to TCP, and hence the overhead. Of course, it also increases the response time, but for noninteractive applications like file transfer, efficiency may be more important than response time to individual requests. Another receiver issue is what to do with out-of-order segments. They can be kept or discarded, at the receiver's discretion. Of course, acknowledgments can be sent only when all the data up to the byte acknowledged have been received. If the receiver gets segments 0, 1, 2, 4, 5, 6, and 7, it can acknowledge everything up to and including the last byte in segment 2. When the sender times out, it then retransmits segment 3. If the receiver has buffered segments 4 through 7, upon receipt of segment 3 it can acknowledge all bytes up to the end of segment 7.
Explained Example: Connection Establishment and Termination
Establishing a Connection
A connection can be established between two machines only if a connection between the two sockets does not exist, both machines agree to the connection, and both machines have adequate TCP resources to service the connection. If any of these conditions are not met, the connection cannot be made. The acceptance of connections can be triggered by an application or a system administration routine.
When a connection is established, it is given certain properties that are valid until the connection is closed. Typically, these will be a precedence value and a security value. These settings are agreed upon by the two applications when the connection is in the process of being established.
In most cases, a connection is expected by two applications, so they issue either active or passive open requests. Figure below shows a flow diagram for a TCP open. The process begins with Machine A's TCP receiving a request for a connection from its ULP, to which it sends an active open primitive to Machine B. The segment that is constructed will have the SYN flag set on (set to 1) and will have a sequence number assigned. The diagram shows this with the notation SYN SEQ 50 indicating that the SYN flag is on and the sequence number (Initial Send Sequence number or ISS) is 50. (Any number could have been chosen.)
The application on Machine B will have issued a passive open instruction to its TCP. When the SYN SEQ 50 segment is received, Machine B's TCP will send an acknowledgment back to Machine A with the sequence number of 51. Machine B will also set an Initial Send Sequence number of its own. The diagram shows this message as ACK 51; SYN 200 indicating that the message is an acknowledgment with sequence number 51, it has the SYN flag set, and has an ISS of 200.
Upon receipt, Machine A sends back its own acknowledgment message with the sequence number set to 201. This is ACK 201 in the diagram. Then, having opened and acknowledged the connection, Machine A and Machine B both send connection open messages through the ULP to the requesting applications.
It is not necessary for the remote machine to have a passive open instruction, as mentioned earlier. In this case, the sending machine provides both the sending and receiving socket numbers, as well as precedence, security, and timeout values. It is common for two applications to request an active open at the same time. This is resolved quite easily, although it does involve a little more network traffic.
Transferring information is straightforward, as shown in Figure below. For each block of data received by Machine A's TCP from the ULP, TCP encapsulates it and sends it to Machine B with an increasing sequence number. After Machine B receives the message, it acknowledges it with a segment acknowledgment that increments the next sequence number (and hence indicates that it received everything up to that sequence number). Figure shows the transfer of only one segment of information - one each way.
The TCP data transport service actually embodies six different subservices:
Full duplex: Enables both ends of a connection to transmit at any time, even simultaneously.
Timeliness: The use of timers ensures that data is transmitted within a reasonable amount of time.
Ordered: Data sent from one application will be received in the same order at the other end. This occurs despite the fact that the datagrams may be received out of order through IP, as TCP reassembles the message in the correct order before passing it up to the higher layers.
Labeled: All connections have an agreed-upon precedence and security value.
Controlled flow: TCP can regulate the flow of information through the use of buffers and window limits.
Error correction: Checksums ensure that data is free of errors (within the checksum algorithm's limits).
To close a connection, one of the TCPs receives a close primitive from the ULP and issues a message with the FIN flag set on. This is shown in Figure 8. In the figure, Machine A's TCP sends the request to close the connection to Machine B with the next sequence number. Machine B will then send back an acknowledgment of the request and its next sequence number. Following this, Machine B sends the close message through its ULP to the application and waits for the application to acknowledge the closure. This step is not strictly necessary; TCP can close the connection without the application's approval, but a well-behaved system would inform the application of the change in state.
After receiving approval to close the connection from the application (or after the request has timed out), Machine B's TCP sends a segment back to Machine A with the FIN flag set. Finally, Machine A acknowledges the closure and the connection is terminated.
An abrupt termination of a connection can happen when one side shuts down the socket. This can be done without any notice to the other machine and without regard to any information in transit between the two. Aside from sudden shutdowns caused by malfunctions or power outages, abrupt termination can be initiated by a user, an application, or a system monitoring routine that judges the connection worthy of termination. The other end of the connection may not realise an abrupt termination has occurred until it attempts to send a message and the timer expires.
To keep track of all the connections, TCP uses a connection table. Each existing connection has an entry in the table that shows information about the end-to-end connection. The layout of the TCP connection table is shown below-
The meaning of each column is as follows:
State: The state of the connection (closed, closing, listening, waiting, and so on).
Local address: The IP address for the connection. When in a listening state, this will set to 0.0.0.0.
Local port: The local port number.
Remote address: The remote's IP address.
Remote port: The port number of the remote connection.
TCP Retransmission and Timeout
We know that the TCP does provide reliable data transfer. But, how does it know when to retransmit the packet already transmitted. It is true that the receiver does acknowledges the received packets with the next expected sequence number. But what if sender does not receive any ACK.
Consider the following two scenarios:
ACK not received: In this case the receiver does transmit the cumulative ACK, but this frame gets lost somewhere in the middle. Sender normally waits for this cumulative ACK before flushing the sent packets from its buffer. But for that it has to develop some mechanism by which the sender can take some action if the ACK is not received for too long time. The mechanism used for this purpose here is the timer. The TCP sets a timer as soon as it transfers the packet. If before the time-out the ACK comes, then the TCP flushes those packets from it’s buffer to create a space. If the ACK does not arrive before the time-out, then in this case the TCP retransmits the packet again. But from where this time-out interval is chosen. Well we will be seeing the procedure to find out this shortly.
Duplicate ACK received: In this case the receiver sends the ACK more than one time to the sender for the same packet received. But, ever guessed how can this happen. Well, such things may happen due to network problem sometimes, but if receiver does receive ACK more than 2-3 times there is some sort of meaning attached to this problem. All this problem starts from the receiver side. Receiver keeps on sending ACK to the received frames. This ACK is of the cumulative nature. It means that the receiver is having a buffer with it. The algorithm used for sending cumulative ACK can depend on amount of buffer area filled or left or it may depend upon the timer. Normally, timer is set so that after specific interval of time, receiver sends the cumulative ACK. But what if the sender rate is very high. In this case the receiver buffer becomes full & after that it looses capacity to store any more packets from the sender side. In this case receiver keeps on sending the duplicate ACK, meaning that the buffer is full and no more packets after that have been accepted. This message helps the sender to control the flow rate.
This whole process makes TCP a adaptive flow control protocol. Means that in case of congestion TCP adapts it’s flow rate. More on this will be presented in the Congestion control topic. Also there is no thing like the negative ACK in the TCP. Above two scenario’s convey the proper message to the sender about the state of the receiver. Let’s now concentrate on how the TCP chooses the time-out-interval.
Choosing the Time out interval:
The timer is chosen based on the time a packet takes to complete a round-trip from a sender to the receiver. This round trip time is called as the RTT. But the conditions i.e. the RTT cannot remain same always. In fact RTT greatly varies with the time. So some average quantity is to be included into the calculation of the time-out interval. The following process is followed.
1. Average RTT is calculated based on the previous results.(Running average)
2. For that particular time RTT is measured and this value depends on the conditions & the congestion in a network at that time.(Measured)
3. To calculate a time out interval:
0.8*(Running avg. ) + (1- 0.8)*(Measured)
The value 0.8 may be changed as required but it has to be less than 1.
4. To arrive at more accurate result this procedure may be repeated many times.
Thus, we have now arrived at the average value a packet takes to make a round trip. In order to choose a time-out interval, this value needs to be multiplied by some factor so as to create some leeway.
Time-out interval = 2*(value arrived in 4th step)
If we go on plotting a graph for the running average and a measured value at that particular time we see that the running average value remains almost constant and the measured value fluctuates more. Below is the graph drawn for both the values. This explains why a running average is multiplied by a value greater than value used for multiplying a measured time.
Comparison: TCP and UDP
The User Datagram Protocol (UDP) and Transmission Control Protocol (TCP) are the “siblings” of the transport layer in the TCP/IP protocol suite. They perform the same role, providing an interface between applications and the data-moving capabilities of the Internet Protocol (IP), but they do it in very different ways. The two protocols thus provide choice to higher-layer protocols, allowing each to select the appropriate one depending on its needs.
Below is the table which helps illustrate the most important basic attributes of both protocols and how they contrast with each other:
The exercise questions here include the assignment questions along with the solutions. This will help students to grab the concept of TCP and would encourage them to go for more exercise questions from the Kurose and the Tanenbaum book.
1) UDP and TCP use 1’s complement for their checksums. Suppose you have the following three 8-bit bytes: 01010101, 01110000, 01001100. What is the 1’s complement of the sum of these 8-bit bytes? (Note that although UDP and TCP use 16-bit words in computing the checksum, for this problem you are being asked to consider 8-bit summands.) Show all work. Why is it that UDP takes the 1’s complement of the sum; that is, why not just use the sum? With the 1’s complement scheme, how does the receiver detect errors? Is it possible that a 1-bit error will go undetected? How about a 2-bit error?
Solution: 01010101 + 01110000 + 11000101 = 110001010
One's complement of 10001010 = Checksum = 01110101.
At the receiver end, the 3 messages and checksum are added together to detect an error. Sum should always contain only binary 1. If the sum contains 0 term, receiver knows that there is an error. Receiver will detect 1-bit error. But this may not always be the case with 2-bit error as two different bits may change but the sum may still be same.
2) Answer true or false to the following questions and briefly justify your answer:
a) With the SR protocol, it is possible for the sender to receive an ACK for a packet that falls outside of its current window.
True. Consider a scenario where a first packet sent by sender doesn't receive ACK as the timer goes down. So it will send the packet again. In that time the ACK of first packet is received. so the sender empties it's buffer and fills buffer with new packect. In the meantime, the ACK of second frame may be received. So ACK can be received even if the packet falls outside the current window.
b) With GBN, it is possible for the sender to receive an ACK for a packet that falls outside of its current window.
True. Same argument provided for (a) holds here.
c) The alternating bit protocol is the same as the SR protocol with a sender and receiver window size of 1.
True. Alternating bit protocol deals with the 0 & 1 as an alternating ACK. Here, the accumulative ACK is not possible as ACK needs to be sent after each packet is received. So SR protocol starts behaving as Alternating bit protocol.
d) The alternating bit protocol is the same as the GBN protocol with a sender and receiver window size of 1.
True. Same argument holds here.
3)Consider the TCP positions for estimating RTT. Suppose that a=0.1 Let sample RTT1 be the most recent sample RTT, Let sample RTT2 be the next most recent sample RTT, and so on.
a) For a given TCP connection, suppose four acknowledgments have been returned with corresponding sample RTTs Sample RTT4, SampleRTT3, SampleRTT2, SampleRTT1. Express EstimatedRTT in terms of four sample RTTs.
b) Generalize your formula for n sample RTTs.
c) For the formula in part (b) let n approach infinity. Comment on why this averaging procedure is called an exponential moving average.
EstimatedRTT1 = SampleRTT1
EstimatedRTT2 = (1-a)EstimatedRTT1 + aSampleRTT2 = (1-a)SampleRTT1 + aSampleRTT2
EstimatedRTT3 = (1-a)EstimatedRTT2 + aSampleRTT3 = (1-a)2SampleRTT1 + (1-a)aSampleRTT2 + aSampleRTT3''
EstimatedRTT4 = (1-a)EstimatedRTT3 + aSampleRTT4 = (1-a)3SampleRTT1 + (1-a)2aSampleRTT2 + (1-a)aSampleRTT3 + aSampleRTT4
EstimatedRTTn = (1-a)(n-1)SampleRTT1 + (1-a)(n-2)aSampleRTT2 + (1-a)(n-3)aSampleRTT3 + . . . (1-a)aSampleRTTn-1 + aSampleRTTn
4) We have seen from text that TCP waits until it has received three duplicate ACKs before performing a fast retransmit. Why do you think that TCP designers chose not to perform a fast retransmit after the first duplicate ACK for a segment is received?
Solution: Suppose a sender sends 3 consecutive packets 1,2 & 3. As soon as a receiver receives 1, it sends ACK for it. Suppose if instead of 2 receiver receives 3 due to reordering. As receiver hasn't received 2, it again sends ACK for 1. So the sender has received 2nd ACK for 1. Still it continues waiting. Now when the receiver receives 2, it sends ACK 2 & then 3. So it is always safe to wait for more than 2 ACK's before re-transmitting packet.
5) Why do you think TCP avoids measuring the SampleRTT for retransmitted segments?
Solution: Let's look at what could wrong if TCP measures SampleRTT for a retransmitted segment. Suppose the source sends packet P1, the timer for P1 expires, and the source then sends P2, a new copy of the same packet. Further suppose the source measures SampleRTT for P2 (the retransmitted packet). Finally suppose that shortly after transmitting P2 an acknowledgment for P1 arrives. The source will mistakenly take this acknowledgment as an acknowledgment for P2 and calculate an incorrect value of SampleRTT.