|
||
 |
||
|
| The Network Framework |
 Email This
 Print ThisView My Personal Library |
|
Protocols February 2001 • Vol.5 Issue 1 Page(s) 36-41 in print issue |
The Network Framework OSI Sets Up The Rules Of PC-To-PC Communication | ||
 In the beginning stages of networking, there was mass confusion. Companies were designing networking systems they felt were the best, but they were also optimizing their systems to work with their hardware. Thus, one company's system would be unlikely to work with another company's hardware. Now, consider that networks consist of a number of different components, not just wiring and what is inside your computer. Clearly, some standards are needed so products from different companies can communicate with one another. The OSI (Open System Interconnection) model is just such a standard.  History. Work on the OSI model (sometimes called the OSI reference model or just reference model) started in the late 1970s, and it came to maturity in the late 1980s and early 1990s. The work on OSI began, largely independently, at two organizations: CCITT (The International Telegraph and Telephone Consultative Committee; CCITT comes from the spelling of the title in French) and ISO (International Organization for Standardization, which is also know as the International Standardization Organization). ISO is a committee of representatives from many different organizations. Its purpose was not to favor a specific method of communication but to develop a standard to let various products interoperate. Both organizations produced working papers about the basic reference model for OSI. Even though both papers descried a seven-layer model, the models were completely different. In 1983, however, the CCITT agreed to adopt the ISO's version of OSI if the ISO made some minor technical changes to accommodate the needs of the CCITT. The ISO agreed and a new standard was born. By 1984, the ISO and CCITT were cooperating to complete OSI and refine it. At the time it was developed, OSI was very radical. Back then, networking was much less common and much less powerful. Hardware vendors locked customers into proprietary systems that forced them to buy all their equipment from one vendor. The goal of the open system was to allow components from different vendors to work together, and that was seen as an invitation to competition, which it was. Naturally, most vendors did not want to invite competition. Given this, it is surprising that the standard was supported widely enough to ever take hold. The OSI model has been very successful at achieving its original goal of letting network components from different vendors seamlessly work together. Presently, the overwhelming majority of commercially available networking hardware and software adheres to the OSI standard (because they have to in order to compete). In fact, OSI has become so successful that the OSI model itself is almost moot. The proprietary systems of the past are gone, and no major customer would consider reverting to a proprietary system so the OSI standards continue uncontested. OSI Layers. To understand OSI, you must understand networks. At its most basic level, a network simply connects equipment together. As simple as that task sounds, in practice it is very complex. Many different things have to be dealt with to facilitate the orderly and efficient transmission and reception of data. Like any complex problem, this problem is best solved by breaking it down into smaller pieces and then solving each piece of the problem individually. The OSI model breaks down the various processes that are needed for two or more computers to interact over a network into seven distinct layers. The specifications for each layer describe how that layer's portion of the process should operate and how that layer will interface with the layers directly above and below it and with the same layer on the other systems to which it is connected. A good analogy is building a house. Overall, a house is a very complex structure. However, when seen as a combination of a foundation, floor, walls, roof, and so on, the structure is much more manageable. Each of those pieces has its own specifications. For example, the specifications might call for 2-x 6-foot boards to be used in the roof trusses. With the specifications for the pieces in place, it is easy to talk about how those pieces fit together and who is responsible for each piece. This also lets a builder plan for the house to have plumbing, for example, without the builder actually understanding how to do the plumbing. All this makes planning and building the house much easier. With OSI, the layers are based on the natural sequence of events that happen during a communications session much as we might specify the pieces of a house based on the construction sequence. From where the data enters the computer to where it reaches the user's application, the seven layers of the OSI model are as follows: 1. Physical 2. Data Link 3. Network 4. Transport 5. Session 6. Presentation 7. Application Layer 1: The Physical Layer. This is the bottom layer. It is responsible for the bit stream (the 0s and 1s that move between machines). For outgoing traffic, it accepts frames (frames are described in the next section) from Layer 2, the Data Link Layer, in the form of bits and transmits those bits serially, one bit at a time. For incoming traffic, it receives the bit stream (the transmission of data in binary form) and passes it on to the Data Link Layer for reframing. Basically, the Physical Layer sends and receives only 0s and 1s. It has no mechanism for determining the meaning of those 0s and 1s. (It is sometimes said to be frame stupid.) The Physical Layer is limited to the processes needed to put the outgoing 0s and 1s onto the transmission media and to pull the incoming 0s and 1s from the transmission media. It does not include the transmission media itself. This transmission media might be coaxial cable, twisted pair wiring, fiber optics, or even a radio signal. Although the Physical Layer does not include the transmission media, it does provide specifications for its performance. The transmission media itself is sometimes referred to as Layer 0. The Physical Layer concerns itself with physical characteristics, such as electrical and optical signaling techniques, for maintaining the physical link between systems. It is concerned with how the data travels. It deals with things such as voltage, current, media time, and impedance levels. Layer 2: The Data Link Layer. To understand the Data Link Layer, you must first understand frames. A frame is a data structure that contains enough information so the data can be sent across a network. Once a frame has been assembled, the network will send it as a single unit. Frames are also called data packets, network packets, or even just packets. A frame contains a mechanism, called a CRC (cyclical redundancy check), that lets the receiver verify the integrity of the frame, which means it makes sure the data transmission is accurate. The CRC is similar to a checksum, which is a numeric value calculated on the bits being sent, but it is calculated using a more complex algorithm that makes it less likely that an error will slip through. When the receiver receives a frame, it must check its integrity and signal the sender that it was received OK or, if there is a problem, signal the sender to resend the frame. The CRC makes this check possible. The sender computes the value based on the bits in the frame. This value is sent with the frame. The receiver also computes this value based on the bits it receives and compares it to the value that was included with the frame. If they match, the frame was sent correctly. If they do not match, there was a transmission error, and a request is sent to resend the frame. The network protocol in use controls the size and structure of the frame. (Network protocols include Ethernet and Token Ring.) Think of a frame as an envelope. Envelopes can have different shapes and sizes and have different contents, but the U.S. Post Office delivers all of them to the addresses on the front of the envelopes. When sending data, the Data Link Layer is responsible for assembling the data and other information into frames and then passing those frames onto the Physical Layer as a bit stream. At a minimum, each frame will contain the following: •A delimiter that marks the start of the frame •The source address •The destination address •The data •The CRC frame check value When receiving data, the Data Link Layer is responsible for rebuilding frames using the binary stream from the Physical Layer. Because the structure of the frame is sent with the frame, the Data Link Layer does not reconstruct the frame but rather buffers (stores) the binary stream until an entire frame is received and checked. In addition to sending and receiving frames, the Data Link Layer handles lost, damaged, and duplicate frames, and it prevents the sender from sending frames too fast. This is known as flow control. Flow control is required because not all systems work at the same speed. Any device on a network, such as a bridge or switch, that is frame-aware is a Data Link Layer device. Bridges and switches use information in the frame header to regulate network traffic. Layer 3: The Network Layer. The Network Layer is a navigation layer. It determines the route to be used between the sending and receiving computer. This is only required when the receiving computer is beyond the local network. A system knows when a computer is off the local network because of its network address. The administrator of a network assigns a network device a network address. It is different from the MAC (media access control) address assigned specifically to each network card. The network address is similar to a ZIP code; its function is to facilitate network communications over long (nonlocal) distances. Routing protocols include IP (Internet protocol), IPX (Internetwork Packet Exchange), and AppleTalk. Layer 4: The Transport Layer. For outgoing data that is too large for one frame, the Transport Layer breaks the data into smaller frame-compatible pieces and assigns each piece a sequence number. These sequence numbers let the receiving Transport Layer reassemble the data into its proper order. The receiving Data Link layer makes sure each frame is correct; the Transport Layer makes sure the frames are reassembled into a complete data set. Because frames are not always received in their proper order, the Transport Layer is also responsible for resequencing as necessary. The Data Link layer requests that corrupted frames be resent; the Transport Layer is responsible for requesting missing frames. Once the receiving Transport Layer has reassembled the data, it is passed to the Session Layer. Layer 5: The Session Layer. The Session Layer manages the connection between two or more computers. This connection is known as a session. The main function of the Session Layer is to establish, manage, and terminate the connections of individual network users. It examines data frames to see that they start, continue, or end a session. It uses this information to determine which computer transmits data and for how long. The Session Layer also makes sure a request for a specific type of service is correctly handled. For example, if your browser requests a Web page, the Session Layer on both systems work together to make sure you receive that Web page and not e-mail. If a system is running more than one network application, say a browser and e-mail program, the Session Layer also keeps the data for these multiple network applications separate and makes sure the incoming data is directed to the appropriate one. The Session Layer is relatively unused as a separate layer. The reason is that many protocols bundle the function of the Session Layer into their Transport Layer. Layer 6: The Presentation Layer. The Presentation Layer makes sure data is received in a format the applications running on the system can use. Not every computer system uses the same data-encoding method. Two data encoding systems that might be used are ASCII (American Standard Code for Information Interchange) and EBCDIC (Extended Binary Coded Decimal Interchange Code.) Data formatting also takes place in the Presentation Layer. This includes character and number formatting. If encryption (encoding) were being used, such as when using a Web browser to make a credit card purchase, the Presentation Layer would encrypt and decrypt the information. If compression were being used to speed transmission, the Presentation Layer would handle it. Layer 7: The Application Layer. The top layer of the OSI model is called the Application Layer. Regardless of what you might think, this does not refer to the applications, such as Netscape or Eudora, you are running on your computer. Rather, it refers to the interface between those applications and the OSI model. For example, when Eudora wants to check for e-mail on the server, Eudora generates a request and hands it off to the Application Layer. The Application Layer then launches a communication session to go out to the mail server and check for new e-mail. Layer Interaction. Even though there are seven distinct layers, these layers do blend together. An OSI model layer typically communicates with three other layers: the layer below it, the layer above it, and its peer layer on the computer with which it is communicating. Even though an OSI model layer functions as though it were communicating directly with its peer layer on the other machine, all communications move down through the layers on the sending machine and up through the layers on the receiving machine. Thus, this peer-to-peer communication is only virtual communication. To accomplish this virtual communication between layers of different machines, each layer of the sending computer adds a header (a section of an electronic message, ordinarily at the beginning of the message, that routes it to its destination and often identifies the sender). Only the same layer, or its counterpart on the receiving machine can recognize and use this header. As the message moves up the layers on the receiving computer, these headers are removed one at a time by the corresponding layers. By the time the message reaches the appropriate receiving layer, the receiving computer has stripped off all the headers and the data is in the same format as it was on the sending machine at the same layer. Making The Model Work. OK, so you understand the different layers and how they interact, but you probably want to understand the whole process in motion. Well, assume you are using Netscape to browse the Web, and it has just requested the Index.html Web page from Smart Computing magazine's server. The Application Layer on your computer would detect that you are requesting a remote file. It would then formulate a request to Smart Computing magazine's server for the Index.html file. Once this request was ready, it would hand off the task to the Presentation Layer. The Presentation Layer would review the request to see if any encryption or data translation was required and, if so, it would perform it. It would then add any information that is needed by the Presentation Layer of the server and pass the packet down to the Sessions Layer. The Sessions Layer would verify exactly what was being requested from the server and would add information to the packet so the remote server would know how to handle the request. It would then pass all this to the Transport Layer. The Transport Layer would ensure it has a reliable connection to Smart Computing magazine's server. It then would break down the information into one or more frames. If it needs more than one frame, it would add sequence numbers to each frame. The frames are then passed on to the Network Layer. The Network Layer takes each frame and adds the address of your computer and the server and passes each frame to the Data Link Layer. The Data Link Layer finishes the process of building the frames. As part of this, it computes the CRC and adds it to each frame. It then converts each frame into 0s and 1s and passes this bit stream to the Physical Layer where they are put onto the network. Depending on the network topology (physical configuration), this could mean waiting for a quiet moment on the network, waiting on a token, or waiting for a specific time. Once this is finished, your browser has successfully sent a request for Index.html to Smart Computing magazine's server. When that server receives the bit stream, the process is reversed and the request is processed. As a result of that processing, these same steps are used by the server to send your browser a copy of Index.html. Once that bit stream reaches your computer, the process is reversed once again in order to feed Index.html into Netscape. As you can see, the seemingly simple process of requesting a Web page to view is really quite complex. OSI Model & TCP/IP. The OSI model competes with the TCP/IP (Transmission Control Protocol/Internet Protocol) as a way to interconnect different brands of proprietary computers and networks across a common global network. When the U.S. government formed a set of specifications for OSI, called GOSIP (Government OSI Profile), it looked like OSI had won the battle against TCP/IP. However, even though the government (specifically the Department of Defense) planned to use OSI instead of TCP/IP, TCP/IP flourished, and it is the basis of the Internet. TCP/IP is similar to OSI, but it has only four layers. (NOTE: THE TCP/IP Application layer performs the same actions as the top three layers in OSI.) 1. Process/Application. This encompasses all the Application and Presentation layers of OSI and part of the Session layer. This layer provides the protocols for remote access and sharing of resources. 2. Host-to-Host. This encompasses part of the Session layer of OSI and the entire Transport layer. This layer provides for connection-oriented data transmission between two or more computers. 3. Internet. This includes part of the Network layer of OSI. This layer is all the protocols necessary for data to travel across multiple networks. 4. Network Access. This includes part of the Network layer of OSI as well as all of the Data Link and Physical layers. This layer is responsible for delivering data over the particular hardware media in use. As for the battle between TCP/IP and OSI, it is now more about bringing the two together. TCP/IP is just too widely used for OSI to completely overtake it. What The OSI Model Is Not. The OSI model provides a conceptual framework for networking computers. However, the OSI model is not a method of communications. By using communications protocols and connection devices, communication is possible. These are rules or guidelines that control how computers exchange information over a network. Network protocols implement one or more layers of the OSI model. In human form, communications protocols represent the language being spoken, say English or Spanish, while network protocols represent how the text is being exchanged, say by book or spoken. What The OSI Model Is. All network devices use the OSI model. Using the OSI model lets Windows, Macintosh, and Unix systems use the same protocols to share data. Most accepted protocols and standards do not fit neatly into one OSI layer. In fact, most common protocols and standards encompass parts of several layers. Vendor implementation of OSI layers is even messier. Vendors implement these accepted protocols and standards, which already mix layers, in different ways. For example, the NCP (NetWare Core Protocol) provides transport, session, and presentation layer functions equivalent to layers 4, 5, and 6 of the OSI model. This is not as much of a problem as it might sound. The OSI model was never intended to have rigid, unbreakable rules. Rather, in implementing the OSI model, it was expected that network vendors would be free to use whichever standard and protocol they deemed to be the most appropriate one. It is, of course, in the vendor's best interest to produce products that conform to the OSI model and work well with other compliant products. If they fail to do so, users will not buy their products. Thus, the OSI model is able to foster open systems without mandating open systems, and that is in the best interest of all users.  by Ronny Richardson View How The OSI Model Works. (NOTE: These pages are PDF [Portable Document Format] files. You will need Adobe Acrobat Reader to view these pages. Download Adobe Acrobat Reader.)
|
|
Home Copyright & Legal Information Privacy Policy Site Map Contact Us