Sie sind auf Seite 1von 7

3.

Network Architecture

Modern computer networks are designed in a highly structured way. To reduce their design
complexity, most networks are organized as a series of layers, each one built upon its predecessor.
The OSI Reference Model is based on a proposal developed by the International Organization for
Standardization (ISO). The model is called ISO OSI (Open Systems Interconnection) Reference
Model because it deals with connecting open systems - that is, systems that are open for
communication with other systems.
The OSI model has seven layers. The principles that were applied to arrive at the seven layers are
as follows:
1. A layer should be created where a different level of abstraction is needed.
2. Each layer should perform a well-defined function.
3. The function of each layer should be chosen with an eye toward defining internationally
standardized protocols.
4. The layer boundaries should be chosen to minimize the information flow across the
interfaces.
5. The number of layers should be large enough that distinct functions need not be thrown
together in the same layer out of necessity, and small enough that the architecture does not
become unwieldy
THE ISO STANDARD MODEL FOR COMMUNICATIONS: OSI
• OSI (Open Systems Interface) Model
• Model = it means that it's only theory! In fact the OSI model is not yet fully implemented in real
networks.
• Open System: It can communicate with any other system that follows the specified standards,
formats, and semantics.
• Protocols give rules that specifiy how the communication parties may communicate.
• Supports two general types of protocols. Both are common.
• Connection-Oriented:
o Sender and receiver first establish a connection, possibly negotiate on a protocol.
(virtual circuit)
o Transmit the stream of data.
o Release the connection when done. E.g. Telephone connection.
• Connectionless
No advance setup is needed. Transmit the message ( datagrams ) when sender is ready. E.g.
surface mail
• Consists of 7 layers: Each layer deals with a specific aspect of communication.
• Each layer provides an interface to the layer above. The set of operations define the service
provided by that layer.
• As a message sent by the top layer is passed on to the next lower layer until the most bottom
layer.
• At each level a header may be prepended to the message. Some layers add both a header and
a trailer.
• The lowest layer transmits the message over the network to the receiving machine. It
communicates with the most bottom layer of the receiver.
• Each layer then strips the header (trailer), handles the message using the protocol provided by
the layer and passes it on to the next higher layer. Finally to the highest layer in the receiver.

(1)

Physical Layer
• Concerned with the transmission of bits.
• How many volts for 0, how many for 1?
• Number of bits of second to be transmitted.
• Two way or one-way transmission
• Standardized protocol dealing with electrical, mechanical and signaling interfaces.
• Many standards have been developed, e.g. RS-232 (for serial communication lines).
• Example: X.21
(2) Data Link Layer
• Handles errors in the physical layer.
• Groups bits into frames and ensures their correct delivery.
• Adds some bits at the beginning and end of each frame plus the checksum.
• Receiver verifies the checksum.
• If the checksum is not correct, it asks for retransmission. (send a control message).
• Consists of two sub layers:
o Logical Link Control (LLC) defines how data is transferred over the cable and provides data
link service to the higher layers.
o Medium Access Control (MAC) defines who can use the network when multiple computers
are trying to access it simultaneously (i.e. Token passing, Ethernet [CSMA/CD]).
(3) Network Layer
• Concerned with the transmission of packets.
• Choose the best path to send a packet ( routing ).
• It may be complex in a large network (e.g. Internet).
• Shortest (distance) route vs. route with least delay.
• Static (long term average) vs. dynamic (current load) routing.
• Two protocols are most widely used.
• X.25
o Connection Oriented
o Public networks, telephone, European PTT
o Send a call request at the outset to the destination
If destination accepts the connection, it sends an connection identifier
• IP (Internet Protocol)
o Connectionless
o Part of Internet protocol suite.
o An IP packet can be sent without a connection being
established.
o Each packet is routed to its destination independently.
(4) Transport Layer
• Network layer does not deal with lost messages.
• Transport layer ensures reliable service.
• Breaks the message (from sessions layer) into smaller packets, assigns sequence number
and sends them.
• Reliable transport connections are built on top of X.25
or IP.
• In case IP, lost packets arriving out of order must be
reordered.
• TCP : (Transport Control Protocol) Internet transport
protocol.
• TCP/IP Widely used for network/transport layer
(UNIX).
• UDP (Universal Datagram Protocol) : Internet
connectionless transport layer protocol.
• Application programs that do not need connection-
oriented protocol generally use UDP.
(5) Sessions Layer
• Just theory! Very few applications use it.
• Enhanced version of transport layer.
• Dialog control, synchronization facilities.
• Rarely supported (Internet suite does not).
(6) Presentation Layer
• Just theory! Very few applications use it.
• Concerned with the semantics of the bits.
• Define records and fields in them.
• Sender can tell the receiver of the format.
• Makes machines with different internal representations to communicate.
• If implemented, the best layer for cryptography.
(7) Application Layer
• Collection of miscellaneous protocols for high level applications
• Electronic mail, file transfer, connecting remote terminals, etc.
• E.g. SMTP, FTP, Telnet, HTTP, etc.
HOW TO MEASURE A NETWORK?
Performance parameters:
Latency Time required to transfer an empty message between relevant computers. Sum total of
• delay introduced by the sender software.
• delay introduced by the receiver software.
• delay in accessing the network.
• delay introduced by the network.
Data transfer rate is the speed at which data can be transferred between sender and receiver in
a network, once transmission has begun. (Bits/sec)
Message transfer time = latency + (length of message) / (Data transfer rate).
Bandwidth is the total volume of traffic that can be transferred across the network.
Max. data rate (bps) = carrier BW · log2 (1 + (signal/noise))
(this maximum is theoretical, not reachable in practice)
Synchronous vs. Asynchronous Communication
Asynchronous
• Data is transferred one char at a time (at random times).
• Each char is surrounded by a start and stop bit.
• Frequently used for terminal communications.
• Hardware logic is simple.
• Problem : If start bit is missed then the whole char is lost.
• Rarely used in high speed communications.
Synchronous
• Data is transmitted in blocks of characters.
• Header and trailer mark the beginning and the end of the block.
• E.g. File transfers from disk.
• More efficient for high speed communication.
• More difficult logic to implement
Simplex vs. Duplex
• Simplex : One way communication all the time. Rarely used.
• Half duplex : Two-way communication, but only in one direction at any time. Common.
• Full duplex : Simultaneous two-way communication. Fast. Common.
Broadband vs. Baseband
Baseband
• Transmission of analog or digital signals without any modulation.
• Uses digital signals to transfer data. (restricted use)
• Not suitable for video and voice.
• Simple, inexpensive, reliable.
• Limited to 1km distance.
• Common in LANs (e.g. Ethernet, Token ring).
OSI INTERNET
Broadband
7. Application Application
• Modulate data before transmission.
6. Presentation • Commonly referred to channel with bandwidth greater
5. Session than 4Khz (voice)
• Need modulation/demodulation devices
4. Transport Transport
• Greater bandwidth and coverage
3. Network Internet
2. Data link
Host-to-network
1. Physical
• More useful for backbones to support more channels

THE REAL STANDARD FOR COMMUNICATIONS: The INTERNET and TCP/IP Reference model
(aka "the Internet suite")
Is the standard de facto for the majority of networks. It has only four layers:
The TCP/IP Model
The TCP/IP model does not exactly match the OSI model. There is no universal agreement
regarding how to describe TCP/IP with a layered model but it is generally agreed that there are
fewer levels than the seven layers of the OSI model. Most descriptions present from three to five
layers. In this technical reference document the layers of the TCP/IP model are defined as follows:
Application Layer: In TCP/IP the Application Layer also includes the OSI Presentation Layer
and Session Layer. In this document an application is any process that occurs above the Transport
Layer. This includes all of the processes that involve user interaction. The application determines
the presentation of the data and controls the session. In TCP/IP the terms socket and port are used
to describe the path over which applications communicate. There are numerous application level
protocols in TCP/IP, including Simple Mail Transfer Protocol (SMTP) and Post Office Protocol
(POP) used for e-mail, Hyper Text Transfer Protocol (HTTP) used for the World-Wide-Web, and
File Transfer Protocol (FTP). Most application level protocols are associated with one or more port
number.
Transport Layer: In TCP/IP there are two Transport Layer protocols. The Transmission Control
Protocol (TCP) guarantees that information is received as it was sent. The User Datagram Protocol
(UDP) performs no end-to-end reliability checks.
Internet Layer: In the OSI Reference Model the Network Layer isolates the upper layer protocols
from the details of the underlying network and manages the connections across the network. The
Internet Protocol (IP) is normally described as the TCP/IP Network Layer. Because of the Inter-
Networking emphasis of TCP/IP this is commonly referred to as the Internet Layer. All upper and
lower layer communications travel through IP as they are passed through the TCP/IP protocol stack.
Network Access Layer: In TCP/IP the Data Link Layer and Physical Layer are normally grouped
together. TCP/IP makes use of existing Data Link and Physical Layer standards rather than
defining its own. Most RFCs that refer to the Data Link Layer describe how IP utilizes existing data
link protocols such as Ethernet, Token Ring, FDDI, HSSI, and ATM. The characteristics of the
hardware that carries the communication signal are typically defined by the Physical Layer. This
describes attributes such as pin configurations, voltage levels, and cable requirements. Examples
of Physical Layer standards are RS-232C, V.35, and IEEE 802.3.
The four layer structure of TCP/IP is built as information is passed down from applications to the
physical network layer. When data is sent, each layer treats all of the information it receives from
the layer above as data and adds control information to the front of that data. This control
information is called a header, and the addition of a header is called encapsulation. When data is
received, the opposite procedure takes place as each layer removes its header before passing the
data to the layer above.

The Suite of TCP/IP Protocols


Unix and the suite of TCP/IP
protocols go hand in hand: it is not
possible to separate the two. TCP/IP
refers to a suite of protocols, not just
the TCP and IP protocols. TCP/IP is
the network portion of Unix. The
following figure relates the Dept. of
Defense (DoD) model of TCP/IP
with the OSI model. The DoD model
is also called the ARPA model (Advanced Research Projects OSI Model and the DoD Model of
TCP/IP Agency). It is not a perfect match
between the OSI Model and the DoD model. However, it is close enough in principle. Note: Only a
few of the major Application layer protocols are displayed;
Network Devices are network interface cards (NIC) and their software drivers. Typically, they are
Ethernet cards, Token Ring cards, and WAN links (such as ISDN or Frame Relay), and they can
also be modems and serial ports. The most common protocol used is Ethernet. It uses an address
burned into the NIC to identify itself to the local network. A typical Ethernet MAC (media access
control) address is a 48 bit number, and would look like 00-02-AF-97-F2-03. Note: the MAC
address is always represented by hexadecimal numbers.

IP stands for Internet Protocol. Its main job is to find the best route--through the Internet--to the
destination. IP uses IP addresses to identify the host machine and the network. A typical IP
address is a 32 bit number, and looks like 142.110.237.1. In this case, 142.110.237.0 identifies the
network address, and 0.0.0.1 identifies the host machine. IP addresses are always represented by
decimal numbers. IP protocol data units (PDUs) are called datagrams, and provide a
connectionless service (send and pray).

ARP stands for Address Resolution Protocol and it is used to map IP addresses to MAC
addresses. This is needed because the Network layer is not aware of the Data Link layer's
addresses (and vice versa).
ICMP stands for Internet Control Message Protocol, and is used mainly for troubleshooting TCP/IP
network connections. Two common programs, ping and traceroute, are part of ICMP.

TCP stands for Transmission Control Protocol, and is used to guarantee end to end delivery of
segments of data, to put out of order segments in order, and to check for transmission errors. TCP
is a connection-oriented service.

UDP stands for User Datagram Protocol, and is a connectionless service. This results in a low
overhead and fast transfer service (relies on the upper layer protocols to provide error checking and
delivery of data).
In the Application layer lies many hundreds of network-aware programs and services such as the
following:

HTTP (80) - HyperText Transport Protocol, which is used for transferring web pages.
SNMP (161/162)- Simple Network Management Protocol, which is used for managing network
devices.
FTP (20/21)- File Transfer Protocol, which is used for transferring files across the network.
TFTP (69)- Trivial File Transfer Protocol, which is a low overhead fast transfer FTP protocol.
SMTP (25)- Simple Mail Tranfer Protocol, which is used for transferring email across the Inter
net.
Telnet (23)- An application for remotely logging into a server across the network.
NNTP (119)- Network News Transfer Protocol, which is used for transferring news.
The numbers, shown in brackets next to the protocols, are called the well-known Port Numbers.
TCP and UDP use these port numbers to indicate where the segments should be sent. For
example, web servers use Port 80 to indicate that the HTTP protocol is used. A Socket is another
name for a well-known Port

Das könnte Ihnen auch gefallen