February 21st, 2003 04:36 PM
The OSI Model (Learn From it)
I feel alot of newbies are lacking basic networking model knowledge.
I found this old tutorial from my college days and thought some of you may find it usefull.
Off course I take NO responsibility for actions or damage caused by this document. All information provided in this tutorial is for educational purpose ONLY.
Well, I've written this tutorial because a friend of mine asked me. I liked that because many people would read it and could learn something. Maybe I'll write some more tutorials later, but I'm reading lots of tutorials and articles myself. So I don't have a lot of time and knowledge to write some good tutorials, but I'll try to do my best... (please accept my poor English). I've written this tutorial especially for newbies who are interested in networking/security and who don't know much about the structure and basics of the Internet and its protocols. If you've already some experience with this subject I suppose you won't learn much from it though. This document explains only the OSI-model in a VERY basic way, so it's easy to understand for newbies. If you want to understand the OSI-model completely you'll have to read some more tutorials, RFC's and books that you can find on the internet.
Have fun... 8)
* The OSI-model
So, what's the OSI-model? Well in the past, communications between computers from different vendors were difficult, because they used different protocols and data-formats. So the International Organization for Standardization (ISO) developed a communications architecture known as the Open System Interconnection (OSI) model that defines standards for linking computers from different vendors. The OSI model has been divided into 7 layers, and the second layer contains also a few sub-layers (divided by the Institute of Electrical and Electronic
Engineers (IEEE) which I won't discuss in this tutorial). Have a look at the following table:
. 7th - layer: Application .> Services .
. 6th - layer: Presentation.> Services .
. 5th - layer: Session .> Communication .
. 4th - layer: Transport .> Communication .
. 3rd - layer: Network .> Communication .
. 2nd - layer: Data-link .> Physical connections.
. 1st - layer: Physical .> Physical connections.
The layers are defined in such a manner so that changes in one layer do not require changes in the other layers. The higher layers (5, 6 and 7) are smarter than the lower layers; the Application layer can handle the same protocols and data-formats used by the other layers, and more. So there is a huge difference between the Physical layer and the Application layer. Maybe this isn't clear, but you'll (hopefully) understand what I mean after you've read the whole document. A request, generated at the top (for example the application Layer) travels
down six layers to the bottom, or physical layer. Maybe you'll think: "Why are there different layers?" Well, the answer is quite simple: Each layer has its own task(s) to perform, let's have a look at the physical layer first:
* The Physical layer (1)
This is the simplest layer; it only has to deal with the electrical (and optical) connections between devices. Binary data is encoded into impulses suitable for transmission over the network medium. For example wires, cables, transceivers & connectors belong all to the Physical layer. Devices like repeaters, hubs & network cards belong also to this layer.
* The Data-link layer (2)
This layer is a little bit "smarter" than the physical layer, because it provides the reliable transfer of data. As the interface between the network medium and the higher-level protocols and layers, the data link layer is responsible for the final packaging of the upper-level binary data into discrete packets before it goes to the physical layer. It sends frames (blocks of data) across a network. Ethernet (802.2 & 802.3), Tokenbus (802.4) and Tokenring (802.5) are protocols at the Data-link layer.
* The Network layer (3)
The primary task of the network layer is to provide the routing functionality by which packets can be sent across the boundaries of the local network segment to a destination that may be located on an adjacent network or on one thousands of miles away. IP, Internet Protocol, is mostly used for this task. Another protocol on the Network layer is IPX, Internet Packet eXchange. The Novell incorporation has programmed this protocol and some others too, like SPX (Sequence Packet Exchange) & NCP (Netware Core Protocol). These protocols have been imported at the Netware Operating System. A few functions that may take place at the network layer:
- The division of the binary data stream into discrete packets of a specified length
- Error detection
- Error correction through the retransmission of bad packets
- Flow control
* The Transport layer (4)
Well, this layer transports data, using protocols like UDP, TCP and/or SPX (Sequence Packet eXchange, this one is also used by NetWare, but it’s in contrast to IPX connection-oriented). The Transport layer is the centre of the OSI-model. This layer provides reliable, transparent transfer of data between end points, this layer provides also multiplexing, flow control and error checking/recovery.
* The TCP/IP protocol-suite and a few other protocols
Before I'll go on with the fifth layer in the OSI-model, I want to explain something about the most common protocols at this moment: TCP/IP. If you already know the basics of TCP/IP it's advisable to skip this part and read further about the session layer.
TCP stands for 'Transmission Control Protocol' and almost all computers are compatible with this protocol. TCP is connection-based, that means it establishes a (constant) connection between two computers and transfers the data between the two hosts. A packet contains a header and a data gram. In the data gram you'll find the data, in the header(s) of a packet is important information mentioned:
- Source Port
- Destination Port
- Sequence number
- Acknowledgement number
- Header Length (Standard 20 Bytes)
- Flags (syn, ack, psh, fin, rst, urg)
- Window size
- IP_v4 or IP_v6
- Header Length
- Total Length
- Flags (Set Fragment bit or not)
- Fragment Offset
- Protocol (this case TCP)
- Header checksum
- Source IP
- Destination IP
If you want to know how a connection is established, please read on...
First a packet with the SYN-flag has to be send to the destination IP, the destination will respond with an ACK(SYN) flag or a packet with the RST-flag. I'll explain: SYN stands for SYN-(chronisation), so it's used to 'ask' the destination computer to establish a connection, when it's accepted, it replies with a packet with the ACK(SYN) flag. ACK stands for ACK-(nowledgement). After receiving the packet with the ACK(SYN) flag, the computer sends back an ACK to tell the other host the connection is established. This is what we call the 'Three-Way-Handshake'. If the connection is established and one of the two hosts wants to disconnect, it sends a packet with the FIN-flag set (FIN stands for FINish). This table will make this (hopefully) a bit clearer:
Computer A Computer B
1) SYN -->
2) <-- ACK(SYN)
3) ACK -->
YES, there's a connection established!
4) FIN -->
5) <-- ACK(FIN)
6) RST -->
Now the connection is closed and the data-transfer(s) are terminated.
Computer A Computer B
1) SYN -->
2) <-- RST
3) ACK -->
****... "Computer B" didn't want to be connected to "Computer A!" (maybe Computer B is running a firewall)
There are a few ways to exploit the TCP protocol: SYN-floods, teardrop, Smurf-attacks, sequence-number attacks & etc. For more information check http://neworder.box.sk/ and http://packetstorm.securify.com/
All right, IP is the abbreviation of Internet Protocol. Look at the example above, if "Computer A" sends a packet to "Computer B" the IP's of both computers are mentioned in the IP-header of the packet, the destination IP (of computer B) and the source IP (of computer A). The routers between the two hosts look in the IP-header and masquerade the packets to the next segment. If an error occurs while a packet is travelling over the (inter)network, for example the destination host is unreachable, ICMP (Internet Control Message Protocol) is used to send an error message so the error could be corrected (and the packet(s) could be retransmitted). The programs 'ping' and traceroute use ICMP messages to detect the status of hosts in other segments.
Currently IP_v4 is standard, but in the future IP_v6, also called IP_ng (next generation), will be used. IP_v4 is based on 32-bits addressing, while IP_v6 supports(!) 128-bits addressing. That's a huge difference, so there're many more IP-addresses available when using IP_v6. The IP_v6 header is a simplification of the IP_v4 header, but it's possible to 'update' IP_v6 (so new features could be added).
* Other important protocols
I've discussed a well known Internet Protocol and Transmission Control Protocol. Let's have a look at a few other protocols (NOTE: they're not all in the TCP/IP-suite):
* UDP, User Datagram Protocol, is in contrast to TCP connectionless. This means a packet travels over the network and reaches the other computer without a connection being established. On its way to the destination the packet could lose because there isn't a direct connection between the two hosts, so UDP is unreliable. UDP is much faster than TCP because there isn't a direct connection.
* FTP (File Transfer Protocol), you probably already know about this one... It is used to upload and download files. Security is handled by requiring the user to specify a user name and password at the daemon (sometimes anonymous login is allowed), but you can use
a packetsniffer to obtain the usernames and passwords. A FTP daemon listens standard at 21/TCP and data travels at 20/TCP.
* HTTP, HyperText Transfer Protocol. HTTP is used to view web pages, in the past a protocol named gopher was used. Gopher can't be used to show nice graphics, only text. So HTTP was developed and it's used by almost everybody. A HTTP daemon listens standard at 80/TCP.
* SMTP, Simple Mail Transfer Protocol is used to send electronic mail (E-mail). Read the sendmail-tutorial at http://blacksun.box.sk/ and you will learn a lot more about this protocol, a SMTP daemon (e.g. Sendmail) listens standard at 25/TCP.
* POP, Post Office Protocol. Nowadays POP3 is the most favourite standard. POP is used to receive and read e-mail from a web page. A POP3 daemon listens standard at 110/TCP.
* DHCP: Dynamic Host Configuration Protocol. If you've got a Cable-connection you use DHCP to get your IP from your ISP.
* OSPF (Open Shortest Path First) & RIP (Routing Information Protocol) are used for determining the best method of routing.
* BOOTP is a protocol used by a network node to determine the IP address of its Ethernet interfaces, in order to affect network booting. The server standard listens at 67/UDP.
* ARP & RARP: Address Resolution Protocol. I'll explain this with an example: If you want to send data to a host and you have its MAC-address (Media Access Control, Ethernet-address) you can query a router for the IP, this is handled by ARP. If you have the IP of a host and you want to know its MAC-address, you can use RARP (Reverse Address Resolution Protocol).
* Telnet (Network terminal protocol) is used for remote logins at computers (usually UNIX-based boxes). Telnet sends the usernames and passwords in plain text over the network, so anybody can obtain them very easily by using a packetsniffer. There's a protocol called SSH (Secured SHell) which is also used for remote logins but it is more secure: The data transferred between the two hosts is encrypted, so the sniffed packets won't reveal any valuable information. Telnet and SSH are usually used in a UNIX-environment. A telnet daemon listens standard at 23/TCP and an SSH daemon listens standard at 22/TCP.
* A typical protocol on UNIX-based systems is finger, nowadays you won't find much boxes with a finger daemon running, standard listening at 79/TCP & 79/UDP. It gives valuable information away (like account-names, and maybe phone numbers which could be used in Social-Engineering attacks). There are a lot of vulnerabilities in the finger daemon so it's very insecure.
Holy ****, this tutorial was about the OSI-model, let's go on quickly with the Session layer...
* The Session layer (5)
The Session layer, because of its name, it is often thought (mistakenly) to be concerned with the network logon procedure and related matters of security. This layer provides service to the upper two layers, it co-ordinates communication between Presentation layer entities. A few protocols at this layer:
NETBIOS: A session interface and protocol, developed by IBM, which also provides services to the presentation and application layers.
NETBEUI, (NETBIOS Extended User Interface), an extension of NETBIOS used in Microsoft networking products, such as Windows NT and LAN Manager.
ADSP (AppleTalk Data Stream Protocol).
PAP (Printer Access Protocol), provides Postscript printer access to AppleTalk networks.
* The Presentation layer (6)
The Presentation layer of the OSI model performs only a single function: the translation of different types of system syntax. For example, a connection between a PC and a mainframe may require a conversion from the EBCDIC character-encoding format to ASCII and many other factors may have to be considered as well. Data compression (and possible encryption) is handled by this layer.
* And finally: The Application layer (7)
This layer is the 'smartest', gateways resides at this layer. Gateways perform the same tasks as a router, but there's a difference between them (look at the following section for more information about this). The Application layer is the primary interface between the application running on a computer and any network resources it requires access to. The Application layer is the only layer a user will operate at, protocols like FTP, telnet, SMTP, HTTP, POP3 resides at this Application layer.
* Routers and Gateways
The difference between routers and gateways is easy to understand when you know something about the OSI-model, and you know about it because you have read the sections above very carefully, didn't you?...
Well, as I told you a gateway is at the Application layer, while a router is active at the Network and Transport layers. This means that a router is faster than a gateway because a gateway has also the protocols at the Session, Presentation and Application layers to route. Hmm, let me make it something clearer with an (stupid) example:
There are two boys doing their homework, the first boy has only English to learn. While the second boy has English and chemistry to learn. The first boy finished his homework faster because he hadn't to do chemistry, but the second boy will be smarter because he did English AND chemistry, so he knows things that the first boy doesn't knows.
Do you understand me? The first boy could be compared with a router (which is faster) while the second boy could be compared with a gateway (which is 'smarter'). Read the example above again if you don't understand it...
February 21st, 2003 04:43 PM
Thanks for another useful tut mark_boyle2002/AO. One of my philosophies is, 'why just limit yourself to just one reference on a particular subject when xx are available'. Reading many views or interpretations on one subject, helps the understandin become more concrete.
February 21st, 2003 04:44 PM
Nice tut. (almost too nice)
Warning. if you didn't write it, give credit.
Here are a couple ways to remember the OSI layers...
Satisfying Tipsy Nuns
All Pimps Say They Never Do Prostitutes.
just like water off a duck\'s back... I AM HERE.
for CMOS help, check out my CMOS tut
February 21st, 2003 04:47 PM
Nice. Keep it up.
Ubuntu-: Means in African : "Im too dumb to use Slackware"