Question

respond to the summaries posted please

Methods for Reliable Data Transfers

• Checksum - Used to detect bit errors in a transmitted packet.
• Timer - Used to timeout or re-transmit a packet if the packet or ACK was lost. Timeouts can happen when a packet is delayed, but not lost (also known as a premature timeout), or when a packet has been received but an ACK is lost.
• Sequence number (seq. no.) - Used for sequential numbering of packets of data flowing from sender to receiver. Gaps in the sequence numbers of received packets allow the receiver to detect a lost packet. Packets with duplicate sequence numbers allow the receiver to detect duplicate copies of a packet.
• Acknowledgement (ack) - Used by the receiver to tell the sender that a packet or set of packets has been received correctly. ACKs may be individual or cumulative, depending on the protocol.
• Negative acknowledgement (nack) - Used by the receiver to tell the sender that a packet has not been received correctly. Nacks will carry the sequence number of the packet that was not received correctly.
• Window (or pipelining) - Sender may be restricted to only send packets within sequence numbers of a given range. By allowing multiple packets to be transmitted, but not yet acked, sender utilization can be increased over a "stop-and-wait" mode of operation.

Answer 1

Transport Layer

The network layer (layer 3 of the OSI stack) is responsible for machine-to-machine communication. The transport layer, one layer higher (layer 4), provides logical communication channels between applications. An application can create an arbitrary number of these channels, each of which has another endpoint on some process running on some host. Writing data onto this channel delivers it to the application that is reading data on the other end of this channel. The transport layer is responsible for implementing this abstraction. Routers in the network are unaware of this concept since they only provide network layer (machine-to-machine) services.

There are multiple transport protocols available on top of IP, including TCP, UDP, and SCTP. TCP and UDP are by far the most popular of these. Two responsibilities of the transport layer are multiplexing and demultiplexing communication channels on the network and, in some cases, implementing reliable data transfer.

Incidentally, a packet at the transport layer is called a segment; it is called a datagram at the network layer and a frame at the datalink layer. We send Ethernet frames, which contain datagrams that are routed by routers. These datagrams, in turn, contain segments that the transport layer of the operating system’s network stack processes.

Transport layer multiplexing and demultiplexing

Multiplexing and demultiplexing are the software mechanisms in place to combine data from multiple logical communication channels on a machine into a single stream of packets on the network and then separate a stream of incoming datagrams into the appropriate communication channels. This is important since communication on multiple sockets shares the same network connection. We can have multiple distinct streams at the transport layer that appear as a single stream of data to the network layer.

Multiplexing is the process of taking data from multiple communication channels (sockets) and sending it out of the machine as a stream of datagrams. Demultiplexing is the opposite process: separating the incoming stream of datagrams into the individual messages for the individual sockets to which each segment is targeted.

The key to IP transport layer multiplexing and demultiplexing is the use of port numbers. Each transport layer segment contains source and destination port numbers. A port number is a 16-bit number that has a unique association to a socket (a communication endpoint) on each host. Naming a socket, also known as binding, is the process of associating a socket with a specific port number and address. The address is the local host’s IP address, of course. In the case where a host has several network interfaces, it will have that many IP addresses and it is possible to make the socket available on only one of these interfaces. More commonly, though, a special address, INADDR_ANY, is used to associate a socket with all available network interfaces. Port numbers are usually specified explicitly for server programs since clients will need to know where to contact them. For example, an SMTP mail server will typically listen for client connections on TCP port 25. A client, on the other hand, will generally not care what port it uses and specifying port 0 is a request for the operating system to pick any available unused port number.