CIS307: Network Architectures

  [OSI Architecture], [TCP/IP Architecture]

Networks encompass a variety of technologies, are created and maintained by large number of ever changing industries, and must satisfy a significant number of often conflicting requirements. As Comer indicates "No single networking technology is best for all needs". A fundamental aim is to support Universal Service, that is, to allow any two computer to communicate, no matter the technologies they use and the specific networks they are directly connected to, as long as there exists a communication path between them. Internetworking is the ability to communicate across networks, with connection between networks provided at the network layer [see below] by routers (a router helps choose a "good" route from source to destination) or, at the data link layer, in the case of networks using the same protocols by bridges, and with possibly different protocols, by switches. An internet is a collection of internetworked networks. The Internet is the name for the global, public internet connecting most networks and using the TCP/IP family of protocols. All the entities connected in a network are called nodes. The computers attached to the network are called end systems, or hosts, or Data Terminal Equipment (DTE). The intermediate nodes of the network are called Intermediate Systems (ISs), or Interface Message Processors (IMPs), or Gateways.

Some significant characteristics of networks are:

• Extent: The physical space covered by the network: Local Area Network (LAN), covering hundred of yards up to a few miles; Metropolitan Area Networks (MAN), covering up to a few tens of miles; and Wide Area Networks (WAN), possibly covering our planet.
• Ownership: Who owns the network: Public, owned by a state licensed entity that will sell services to users; and Private, owned fully by a company for its private use (it may use cables leased by a service provider).
• Service: The way information is exchanged between users: connection-oriented and connectionless. In the former case a connection is established between the communicating agents and on that connection messages are exchanged (the connection can be persistent, i.e. fixed usually in hardware, and switched, i.e. that can be set/reset rapidly with commands.). In the latter instead each message is treated as a free standing entity (datagram). When a connection is established, the connection receives an id or address that is used for routing the packets involved in the connection. Related concepts at the communication level are circuit switched and packet switched. In the former the circuit is set at the wire level [the wire can be multiplexed among circuits using FDM or TDM] and communication between interlocutors always follows the same path. In the latter, individual packets follow their own independent routes, or, if a virtual circuit is set for the duration of the connection, they follow the same virtual circuit route.
• Quality of Service: Here we refer to things like delays, reliability, jitter,  and throughput. Delays can be due to various causes, we have propagation delay (time for transfers "on the wire"), switching delay (time to move across switches, bridges, routers), queueing delay (time waiting queued at a node waiting for a transmission channel to become available), access delay (time to wait to insert a message, as in Ethernet, onto a shared medium). Jitter,  that is the variation in the duration of delays, such variation is or not significant depending if we are sending data in a backup (it is not), or in a real time-application (it is), or we transmit voice or video (it is). Throughput, that is the data rate at which information is actually transmitted. Reliability: that is, the availability of service and the infrequency/absence of errors.

Both the the OSI and the TCP/IP architectures are layered architectures, that is the functionality of the network is decomposed into layers, where a higher level layer uses the services provided by the layer immediately below it and where across a network communicating entities communicate exclusively at the same layer (i.e. if a sender entity at layer i sends a packet P, the receiver entity at layer i will receive P; these two are called peer entities).

OSI Architecture

   application -->    DNS      SMTP      FTP    TELNET    HTTP
                       |         |        |        |        |
                       |         |        |        |        |
                       |         |        |        |        |
                       |         +--------------------------+
                       |                              |
                       |                              |
   transport ---->    UDP                            TCP
                       |                              |
                       |             ICMP             |
                       |              |               |
                       +------------------------------+
                                      |
                                      |
   network ------>                   IP<--------------+
                                      |               |
                      ARP             |               |
                       |              |               |
                       +--------------+               |
                                      |               |
    link   ------>                 Ethernet          PPP
                                      |               |
                                      |               |
    physical ---->                 Ethernet         modem

There is a fundamental difference between the lowest three levels (the communication subnet) and the top four levels of the OS architecture. The bottom layers are between directly connected hosts thus they involve all the hosts in a path from sender to receiver. The top four layers are end-to-end protocols, that is, the communication is stated in terms of only the original sender and the final destination, independent of how many intermediate hosts are traversed. Intermediate nodes do not participate at all in the processing of the higher level protocols, to them it is data. [Think in terms of overhead: In the source and target node protocols at all layers are processed. In the intermediate nodes only protocols in the bottom three layers are processed.] This has a direct impact on efficiency: for example, error checking in protocols at the higher level is only done at the sender and receiver, not at each intermediate hosts.

Three related concepts in discussing layers are Services, Interfaces, and Protocols.
Services are what a layer provides to the layer above it (the former is the Service Provider and the latter is the Service User). [A layer may provide more than one service. For example the transport layer provides a connection-oriented service and a connectionless service.] A service will be expressed as a set of primitives used to access the service. There are four basic kinds of primitives:

• Request: A requesting entity asks for a service.
• Indication: A responding entity receives a request for service.
• Response: A responding entity responds to a request.
• Confirm: A requesting entity receives a response to a prior request.

Each message (message is the preferred name for the PDU at the session layer and above; at the transport layer segment is preferred; at the network layer datagram and packet are used; at the data link, frame is used; and at the physical layer packet is common) consists of data being transmitted plus information required by the protocol for addressing, error detection, etc. This extra information appears as a header before that data and (may be) a trailer after the data, i.e. the data is encapsulated in the message. [Not all messages have both a header and a trailer. Usually the trailer is not present.] The message sent at layer i will be transmitted as data by the layer below it. Assuming that the layer below can transmit this data as a single message we will have the situation

Note that the headers and tails constitute transmission overhead, reducing the utilization of the bandwidth of the communication channel. Of course this is only part of the communication overhead: retransmissions and acknowledgements further reduce bandwidth.

A concept often used in communication is session. It consists of one or more connections. For example, a program on machine A may be involved in communication with a program on machine B, the connection drops due to communication problems. When communication is reestablished the programs continue from where they were in the session using a new connection. An example of a session is the interactions during a remote procedure call. The requestor has to send the request, possibly in multiple messages, the receiver collects them, processes the request, then sends back the results. All in one session.

The transmission units at different layers may be of different maximum sizes. We have here the same distinction that exists between "logical" and "physical" records in file systems.

Routing is the process by which a message moves from a sender to a receiver across multiple intermediate hosts. It is a very complex activity. Part of the complexity is that we want to have a system that keeps on working while hosts and links fail and come on line at their own pace. Routers are placed between neighboring networks to decide which of the incoming packets should be kept on the local network and which should me moved to the other network. It is not an optimal process: packets keep a hop count that is increment with each transfer. When it gets too big, as in a loop, the packet is discarded. A router maintains a routing table that, among other things, specifies for a message with a given destination, to what output link to forward it to. We will talk more about routers and routing.

A number of issues should be kept in mind when we analyze a protocol and its implementation. Among them:

• Framing: How do we recognize the beginning and end of a message (or packet, or frame, or ..);
• Flow Control: How do we make sure we don't send too much information to the receiver so that it cannot handle it and has to "drop it on the floor";
• Congestion Control: How do we make sure that we don't have too many messages trying to get into a communication channel. Thus causing delays and perhaps lost packets;
• Multiplexing: How a single communication channel can be shared for more than one conversation;
• Addressing: How do interlocutors address each other [host interfaces are known by their IP address (a 32 bit integer); unfortunately in ethernet the address understood by the hardware is the Ethernet address (a 48 bit integer) of the connector; thus there is need of a way to translate from IP to Ethernet addresses. In general, this requires an address resolution protocol (ARP is such a protocol).
• Error Detection and Correction: what we do to detect if an error has occurred in transmission and how to correct it.
• Fragmentation: One message at a layer may have to be fragmented over multiple messages at a lower layer. A layer may be able to accept messages of length N, but the implementation of the layer in terms of a protocol at the layer below may accept only messages of length M, with M < N. Thus the original layer must be able to fragment the original message into a number of messages at the lower level. These fragment may then be sent across different routes and the receiver will have to worry about missing fragments, incorrect received order, in recombining them into the correct received message. This problem occurs in particular between the network layer (IP protocol) and the data link layer, where the IP packet needs to be split into a number of fragments.
  The opposite situation may also occur: At the data link layer with Ethernet we have messages of up to 1526 bytes. At the network layer we may have a limit of 128 bytes. [Or the data link unit may be 53 bytes...]

TCP/IP Architecture