Table of Contents Chapter 5 Chapter 7 Netguru.net Glossary
Chapter 6 - Networking
Standards and
References
Standards play an important role in networking. Without standards, manufacturers of networking products have no common ground on which to build their systems. Interconnecting products from various vendors would be difficult, if not impossible.
Standardization can make or break networking products. These days many vendors are hesitant to support new technology unless there is a standardization base from which to work. Vendors want to know there will be some measure of interoperability for their hardware and software. Otherwise, releasing a product could be a marketing nightmare if it is not compatible with standards that are later embraced by the marketplace.
Reference and Standard-Setting Organizations
There are several sources for standards. Vendors may provide standards and references. Anybody who ever purchased in IBM clone can testify to that. Also standards may be created by organizations devoted to setting them up. Among the most well known are the International Organization for Standardization (called by the old acronym of ISO) and the Institute of Electrical and Electronics Engineers (IEEE or "I triple-E").
1. International Organization for Standardization (ISO)
The ISO was founded in 1946 and is currently headquartered in Geneva, Switzerland. Its mission at its inception was to create international standards regarding the threads of screws used for manufacturing world-wide. As needs for other standards arose, the ISO (then called the International Standards Organization, hence, ISO) stepped in to handle the task, and their influence crept into all areas of manufacturing and services.
The primary focus for the ISO hasn't really been in the electrical and electronics area. Many of the standards set up in those areas were created by an older standards-setting organization called the International Electrotechnical Commission, or IEC, also based in Geneva. However, the ISO has responded to the needs of computing standards by forming a joint committee with the IEC dealing with information technology. The ISO has published networking standards even though they did not necessarily create them but borrowed them from other sources. An example is the 802 series of standards developed by IEEE and reissued by the ISO as the ISO 8802 protocols. These deal with subjects such as Ethernet LANs and token ring LANs.
The International Organization for Standardization is made up of over 160 technical committees with over 2,300 subcommittees across the globe. Most of these committees work with national standards organizations from several countries. All told, there are over 75 of these national groups.
In the US, the standards-setting body and member of the ISO is the American National Standards Institute (ANSI - pronounced an-see). ANSI is headquartered in New York City. It has been around since 1918 providing voluntary standards for both the private and public sectors. It has been instrumental is assisting in the adoption of industry standards dealing with many areas of information technology, including everything from programming languages to disk drives.
Perhaps the most notable contribution that the ISO has provided to networking is the OSI model (Open Systems Interconnection). The OSI model basically details all the functions of networking and provides a framework in which all vendors around the world can create systems that can communicate with one another. All networking vendors to some degree have adopted and supported the OSI standards. The OSI model is discussed at length in this chapter.
2. Institute of Electrical and Electronic Engineers (IEEE)
The IEEE has done notable work in the standards area of networking. This organization is huge with over 300,000 members made up of engineers, technicians, scientists, and students in related areas. The Computer Society of IEEE alone has over 100,000 members. IEEE is credited with having provided definitive standards in local area networking. These standards fall under a group of standards known as the 802 Project executed by the Computer Society's 802 subcommittee.
The 802 standards were the culmination of work performed by the subcommittee starting in 1980. The first published work was 802.1 which specified a framework for LANs and internetworking. This was followed in 1985 with specific LAN-oriented standards titled 802.2 - 802.5. Since that time there have been other references set up as well. Most of the work performed by the 802 Project committee revolves around the first two layers of the OSI model initiated by the ISO. These layers involve the physical medium on which we move data (cable type) and the way that we interact with it. It addresses such crucial issues of how data is placed on the network and how we insure its accuracy and flow. In order to better define these functions, the IEEE split the Data Link layer of the OSI model up into two separate components. This is illustrated in Figure 6.1.
Here is a summary of what committees there are and what standards areas are being defined within IEEE:
802 IEEE committee responsible for setting standards concerning cabling, physical topologies, logical topologies and physical access methods for networking products. The Computer Society of IEEE's 802 Project Committee is divided into several subcommittees that deal with specific standards in these general areas. Specifically the Physical layer and the Data Link layer of the ISO's OSI model are addressed.
802.1 This work defines an overall picture of LANs and connectivity.
802.1B This set of standards specifically addressed network management.
802.1D Standards for bridges used to connect various types of LANs together were set up with 802.1D.
802.2 Called the Logical Link Control (LLC) standards, this specification governs the communication of packets of information from one device to another on a network. Specifically it deals with communication, not access to the network itself.
802.3 Defines the way data has access to a network for multiple topology systems using Carrier Sense Multiple Access/ Collision Detection (CSMA/CD). A prime example is Ethernet and StarLAN systems. These LAN types operate at 10 Mb/sec.
802.4 Standards developed for a token-passing scheme on a bus topology. The primary utilizer of this specification was the Manufacturing Automation Protocol LANs developed by General Motors. Operates at 10 Mb/sec.
802.5 This standard defines token ring systems. It involves the token-passing concept on a ring topology with twisted pair cabling. IBM's token ring system uses this specification. The speed is either 4 Mb/sec or 16 Mb/sec.
802.6 Metropolitan Area Networks are defined by this group. MANs are networks that are larger than LANs typically falling within 50 kilometers. They operate at speeds ranging from 1 Mb/sec up to about 200 Mb/sec.
802.7 These are standards concerning broadband LANs.
802.8 This group sets up standards for LANs using fiber optic cabling and access methods.
802.9 This specification covers voice and digital data integration.
802.10 These members set standards for interoperable security.
802.11 Wireless LANs are the subject of this particular subcommittee's works. Both infrared and radio LANs are covered.
There are groups within these groups defining more of the specifics of each of these topic areas. Many of these broad topic areas are further defined. A case in point is the different ways in which Ethernet can be used. Ethernet LANs may use twisted pair and star topologies. It may use coax cable with a bus topology. It may use fiber. These are all defined within the 802.3 area and are referred to with names such as 10BASE-T, 10BASE2, 10BASE5 and 10BASE-F. These are covered in more detail in the next chapter.
The Open Systems Interconnection Model
Figure 6.1: The Open Systems Interconnection model has seven layers that describe functions of data communication for networking.
Perhaps no other standard has affected networking more than the OSI model. Presented by the ISO in the late 1970s, this model was to serve as a framework for worldwide communications. It has been adhered to in one respect or another by all network vendors. However, few have based their own implementation completely on the model with its seven layers of functions. Most believe that having individual functions broken into so many layers is impractical for their protocols (packet types) because of the overhead each layer adds. This will become clearer to you as we examine the functions of each layer and what's involved in using those functions.
If all the world used one type of computer, our lives would be so much simpler. Unfortunately, a reality check tells us that not only will we see a great variance in the types of computers from one company to the next, but we are likely to see a wide variance from one office in our company to the next. Hardware differs depending on whom we purchase our machines from and what purposes we intend for it. One group might need a mainframe, while another might require a couple of PCs. As need for information grows, we eventually reach the point where we want those different machines to communicate. It's times like these that the OSI model proves useful.
Every machine that can be connected to a network goes through similar process in transferring that data out on the wire. An application that we are running on that device generates some data that it wants to send to some other entity on the net. The information must be placed in a format suitable for the application that will receive it on the other side. Once this is done, the machine goes through the process of encoding the data into a network-ready format. This is done by breaking the data up into small units called packets. The packet not only contains raw data (just a few bytes in each packet), but it contains other important information such as where the data will go. We'll spend more time on the actual contents of several types of packets in the next couple of chapters.
As the data is being prepped for transfer it is, in effect, passing down through the layers of the OSI model. The highest layer is the application, the lowest is the cable or other physical medium. While passing through these layers, other information may be tacked on to the packet in order to ensure the data is delivered correctly. Once the information is received by the recipient machine, the data passes up through the layers where information that has been tacked on at the sender is peeled off. Last on the layer list is the application running on the receiver device. It gets the raw data originally sent by the source machine. Figure 6.2 illustrates this concept.
Let's take closer look at what the various layers of functions are:
That's a rather brief explanation of the functions of each layer. However, it is sufficient for you to be able to grasp each layer's function. We'll provide any other information concerning the layers' functions as needed.
Data Frame Construction and Use
As data is making its way through the successive layers of the OSI model prior to sending, pieces of information are tacked on that will be useful in getting the data to the right layer on the receiving end. Let's look at an example of how and when this is accomplished.
Let's say that you decide to send an electronic mail message to a co-worker asking that person to call you. You create a message, "Call me." The application at the application layer appends what as known as a "header" (represented by AH, PH, etc., in Figure 6.2) to your message to identify what kind of application this message is for (E-mail application). Then the message passes down to the presentation layer.
The presentation layer takes both the data from the application and application header and groups it together. This group is known as a "data unit". On to this data unit, the presentation layer may add its own presentation header before passing it down to the session, transport and network layers. Each of these may add their own headers as necessary. Every time a header is added, it is grouped with the information that it has been appended to, and the whole collection becomes a data unit for the next layer.
Eventually the data reaches the data link layer. Here the data is grouped into frames by placing framing information before and after each group of a predetermined number of bits or bytes. In addition, address information is added, which basically tells where the frame is going (destination) and where it is coming from (source). Plus, control information is added as well as the Frame Check Sequence (FCS). The FCS is used for error detection. It is a 32-bit value created by putting values from other fields in the frame through a polynomial equation. The result is unique to that frame. When the frame is received the same process of using the polynomial equation is repeated to determine if the data in the frame has gotten corrupted. This method is extremely reliable. The chance for a corrupted frame slipping through this error-checking process is one in four billion.
The next step is to move the data down to the physical layer. At this level the media, whatever it is, is accessed. The streams of data bits are placed on it and its serial communication from one device to another is monitored.
Eventually, the data bits get to the other machine. Here the entire process is reversed. The streams of bits coming in from the physical layer are checked by the data link layer. If there is a problem, then the upper layers can be informed of it. The data link layer can request a packet to be re-sent if it didn't pass the frame check sequence test. The header and trailer information placed on the data at the source is stripped away and the data eventually reaches the application layer on the receiving machine where your co-worker is. The program that handles E-mail then intercepts the message and informs your co-worker to call you.
Another way of thinking about data frames is to think of an onion. Each layer of the OSI model may add an additional layer to your "onion" (or application data). When the packets are received, the data is reconstructed by peeling the onion.
Figure 6.2: Information is added to data as it descends through the OSI layers. When received the data will be reconstructed by removing the information added.
Important IEEE Specifications
As mentioned earlier, the IEEE has provided many useful specifications. Several of these are discussed in the next chapter which deals with popular network types. We'll start by introducing a sampling of the 802 Project Subcommittees' works. Each subcommittee deals with specific functions of the OSI model.
1. 802.1 - Higher Layers and Management
Earlier, we introduced you to bridges that were devices that allowed networks to connect together at the data link layer. The 802.1 committee is responsible for providing specifications for bridges. So far, the committee has given standards for "Spanning Tree Bridges", which are those which are implemented currently with Ethernet systems. The 802.1D subcommittee is working with "Source Routing Bridges" which are an IBM offering. This type of bridging is used with token ring systems.
In addition, 802.1A is responsible for adopting a network management specification that is consistent with the OSI model. As IEEE has spent a great deal of time and effort defining standards around this model, it makes since to focus management tools around it as well.
The 802.1B subcommittee develops network management protocols. Currently there are a few competing protocols. 802.1B attempts to keep these in order and stabilize the network management picture.
2. 802.2 - Logical Link Control
One of the first things that this subcommittee did was to divide the traditional data link layer of the OSI model into two separate layers. The resultant Logical Link Control layer and Media Access Control layer (MAC layer) made life much easier for the network designers by adding flexibility. At this point, only the MAC layer is dependent on what protocols you are using (i.e. Ethernet, token ring). The LLC layer functions independently providing a pathway for data to flow to the upper layers of the OSI model without those layers having to worry what kind of network you are using.
The Logical Link Control layer's main function is to make sure that communication takes place on the network with no errors. Basically, It has to report to the bosses in the upper layers and it would just as soon not have to report any problems. The communication processes involve error correction, acknowledgments for receipt of information, creating of connections between network devices and the tearing down of those connections, and the ability to number (or sequence) each packet. Some of the services provided by LLC overlap those of the transport layer of the OSI model. Therefore, if those services were not provided by the transport layer, they may be available through LLC.
The LLC layer communicates with higher layers via Service Access Points. When, for instance, the network layer wants to pass a data unit down to the data link layer, it requests the data link layer to accept the data and continue preparing for transmission at a Service Access Point (SAP). When data is traveling up through the layers, the SAP allows the LLC layer to request that the network layer take the data and remove the network header (NH in Figure 6.2). Several processes from the higher layers may be requested. Each process will have a unique SAP address. In this manner, the MAC layer that is discussed below (which has only one address as far as the network is concerned) can communicate with several higher layer processes.
Figure 6.3: Service Access Points (SAPs) provide a way for lower layer processes to communicate with higher layer processes.
The other layer of the IEEE's data link layer is the Media Access layer (MAC). Here data is placed in the proper format for the type of network you are using. This layer is in charge of providing source and destination addresses, error detection and grouping of data into frames.
When data descends to the LLC layer, it is divided into frames. A LLC frame consists of several components that together are referred to as a Protocol Data Unit (PDU). There are three kinds of these PDUs. One carries information in a data transfer, another supervises that transfer, and another creates or destroys the communication.
Figure 6.4: The LLC frame contains several fields of data.
The LLC layer frame begins with a specific SAP field specifying what process is requested by the sender. This field is known as the Destination Service Access Point field (DSAP). It is 8 bits in size. When the frame passes down to the MAC layer below it prior to transmission on the wire, it acquires a MAC header that directs the frame to a particular node. Plus, the frame gets a Cyclical Redundancy Check (CRC) field that is used for determining if the data has any errors.
The DSAP is followed by the Source Service Access Point field (SSAP) that informs the recipient what process at the sender is communicating with the recipient's process in DSAP.
Next a control field that is used for various purposes depending on the processes uses up 8 or 16 bits. This field is what determines which type of PDU the frame is. It is also used for keeping frames sequenced in the event that frame sequencing is used.
Finally, we have the information that has been passed down from the layers above the LLC layer. The amount of bits in this field may be determined by the type of network being utilized.
Once again, verification of the packet information takes place on the MAC level so the LLC layer doesn't have to re-perform this testing. The LLC layer is responsible for helping to correct errors. This is done in a variety of ways. The LLC layer can send acknowledgments from a receiver to sender to ensure data was received. Obviously, failure to receive an acknowledgment may mean data was lost so the sending entity knows to send again.
The LLC layer also has the ability to place sequence numbers in each packet so that packets received can be properly ordered. This also prevents any errors and the receiver knows if certain packets have not been received. In addition, if a relationship (connection) is established between sender and receiver, the LLC layer can monitor the connection. If there is a momentary failure of the entities to communicate, the LLC layer can reset the connection and allow the transfer of data to continue. Depending on the circumstances, some data could be lost.
The LLC layer also provides protection against errors using flow control in which the sender is informed if it trying to send data too fast. There are several methods of flow control. Most are similar to what we see in everyday modem communications.
Previously we had mentioned that three types of PDUs exist for use in the LLC layer. The Control field of the LLC frame contains information that determines which type of PDU is being used. Each type of PDU is used to provide a different kind of service. Let's look at the services provided by LLC.
1. Connectionless, Unacknowledged Service means that there are no special relationships set up between sender and receiver. Plus, there are no acknowledgments sent beck from receiver to sender. This service involves very little overhead, is very fast, and is least reliable.
2. Connection-Oriented, Acknowledged Service means that a relationship is set up between sender and receiver. They agree on parameters for communication. There are acknowledgments sent back from receiver to sender to ensure flow control and error checking. This service provides the slowest performance due to overhead, but the most reliable means for delivery.
3. Connectionless, Acknowledged Service means that there is no special relationship set up between sender and receiver, but acknowledgments are traded yielding flow control. This is the best of both worlds from the other service types. It has some overhead, but has reliability features as well.
These types of services are grouped into what is known as LLC Service Classes. The classes differ in which services they contain.
LLC Classes of Service
It would probably worth our while to discuss the flow control mechanisms of LLC in more detail. When we send data serially, we must have some method of checking the validity of the data. Usually this involves calculating a number based on the number of 1s or in a block of data and placing the calculation result onto the data block. We call it the Cyclical Redundancy Check or CRC. The same calculation is performed at the receiving end, and if there is a discrepancy, a retransmission is requested of the sender. In order to let the sender know the CRC test was successful, an acknowledgment is sent back from the receiver.
Traditionally the sender had to stop broadcasting while waiting on a positive acknowledgment from the recipient. The technical term for this is positive acknowledgment with retransmission. This refers to the acknowledgments and the what happens if an acknowledgment is not sent - retransmission. Novell calls this feature stop-and-wait technology, so be aware of the difference in terminology.
Since data can only be traveling from sender to receiver or receiver to sender (acknowledgments) at one time, this is inefficient use of network resources. The solution is to be able to send a packet without having to wait for an acknowledgment for the previous packet. This methodology is called "sliding window technology".
The basic premise behind sliding windows is that we are busy sending instead of waiting. For example, let's say we have nine packets to be sent. We would transmit packet #1 and then go on to #2 without waiting for the acknowledgment (abbreviated ACK) for #1. We continue to send packets until we reach what we have pre-determined to be our window size at #6. At this point we do receive an ACK back for #1. The packets that have been sent but are as yet unacknowledged, are said to be "in the window". At the point that #1 ACK is received, the window slides up and the #7 packet is released. In this manner, we always have packets going out ahead of the acknowledgments. Figure 6.5 illustrates this for you graphically.
Figure 6.5: The standard stop-and wait method as compared to the sliding window technology yields much less throughput across a network.
The window size can be adjusted for the network. Obviously, a small window means that there is potential waiting for ACKs. A fast network could potentially accept a large amount of packets, so a window might be larger in such cases. In every instance, both the sender and receiver carefully keep up with what packets have been sent and acknowledged.
In 802.2 implementation, LLC Service Type 2 uses sliding widows flow control where LLC Service Type 3 uses what Novell calls stop-and-wait.
Chapter 6 Study Tips
1. Know what major standards-setting organization there are.
2. Know what the leading contribution the ISO has made to modern networking.
3. Understand what ANSI's relationship is to ISO.
4. Know what IEEE stands for and what the organization does.
5. Be able to briefly describe what topic area each 802 committee deals with and the appropriate 802 number for that committee (i.e. 802.3).
6. Know the seven layers of the OSI model and what functions take place on each layer.
7. Know how the IEEE 802.2 committee altered the OSI data link layer.
8. Know how data is processed through the OSI layers including the addition of headers and trailers.
9. Know what a Service Access point (SAP) is and how it functions.
10. Know the functions of the 802.2 LLC layer and MAC layer.
11. Know the contents of an LLC frame and what each field does.
12. Explain the terms "connectionless", "connection-oriented", "acknowledged", "unacknowledged", "sequence number".
13. Know the LLC classes of service.
14. Explain the function and advantages of "sliding window" technology.
Table of Contents Chapter 5 Chapter 7 Netguru.net Glossary
Copyright© 1993 & 1997 by Atrium Technical Inc. All rights reserved. Illegal to copy without written permission of ATI.