Networking Standards

Packets in Data Transmission

Let's look at data transmission in computer networks and specificallythe packets that we are using for that. So in modern computers data are usuallystored in files, and this files can be small, large, or veryvery large. So one might assume that when we transfera file from one computer to another the whole file is just send as acontinuous stream of ones and zeros over telecommunication networks.And this would be the best way to do it may be an ideal network,error-free with infinite communication capabilities.But real networks are not like that, obviously, they haveerrors and have finite communication capacity which is calledbandwidth. So we need to deal with that.So if computer sends a large amount of data over a networkwithout any pause it will cause the other computers to waitwhile the data is being moved and this would disturb the communicationof the rest of the users. And then think of when what happens when there'san error then a large amount of data has to be retransmittedover the network which would just add up to more traffic in the network.So to minimize these effects what we do iswe split large data units and reformat them into smaller packages,because this allows for better network sharing and better managementof errors in the transmission. So this makes it also possible to mixpackets from different users if an error occurs.So then if an error occurs you only have to re-send a small sectionof the data and not the whole big file again.And then it's relatively easy to recover from this error.So in order for many users to transmit at once data quicklyand access network we break data into small manageable chunksand these chunks are called packets or frames.Although the terms packets or frames are often used interchangeablythere are some differences, this is because the communicationprocess on the computer network is organized in layersand the word frame is usually used to name the dataaggrupation in the lower layers and the word packet is used in layersthat sit above them.So a packet or frame needs a way to marks its beginning and itsending and it needs to carry some control information.So different communication protocols use different conventionsfor distinguishing between the elements of the packets and for fotmattingthe data. So for example, in a lot of point two pointprotocols the packet is structured in eightbits bytes and special characters are use to delimit different elements.And there's other protocols, like for example the Ethernet. Andthey use special sequence of bytes to mark the beginning and theending of frames. So these are just some protocols that organizedtheir packets as bytes but this is not always like that, there's alsoprotocols that defines their packets at bit level insteadof byte level.But what most of the packets and frames have isa header, a body and a footer. So if we look at the IP packets forexample, we can see that in the header we have information aboutthe destination address, the ip address of the packet.There's information about the length of the packet,there is information of the originating ip address,the header also includes the packet number in the sequence of packetsand establishes the protocol to determine the type of packetsthat iss being transmitted so let's say audio or video, email, web pagesetc and the header contains information to perform any errorchecking operation and the packet body which is also calledpayload contains the actual message is being sentand packet footer that is optional contains a sequenceof bits that indicates the end of the message and it also maycontain data and information that we need to perform some error checking operations,for example at the receiving end.So error detection and correction is performed at variouslayers in the protocol stack and network packages for that containthings called checksums or parity bits or cyclic redundancychecks. And all these are mechanism devised to be able to detectand correct errors that occurred during the transmissionof a packet by operating with the transmitted data.So at the transmitters side a checksum or parity bit or a cyclicredundancy check code is calculated before the packet is beingsent. So this code is transmitted with the data and when thisdata is received at the destination the checksum is recalculatedand compared with the one in the packet. If discrepancies are found so it's not thesame, then it means that the packet may be corrected or discarded.And any packet loss isdealt with the network protocol. So simple protocols, simple codesallow only for detection of errors, while more complex codes cancorrect up to a certain number of errors.And in many cases modification of the network packets may be necessarywhile routing and in this case checksums are just being recalculated.Under fault conditions packets can end up following closed circuits.And if nothing is done to avoid it the number of packets recirculatingcould buid up until network is congested, for exampleat the point where it will actually fail because of congestion.The time to life is a field that is in a packet and it's decreasedeach time a packet goes through a network node and if the field reacheszero routing has failed, it's assumed and the packet is discarded.And this way it doesn't fill up the network.Ethernet packets do not hava atime to live fieldand so they are actually subject to what is called a broadcast stormin the presence of, for example, a switch loop.So we have look a little bit at packets in data transmission in computer networks.


Ethernet Discussion

In this unit were going to see the technology used for local areacommunications, ethernet. Ethernet was created aroundnineteen seventy three at Xerox as a network standardto communicate several computers using a shared cable.To access this shared medium it uses a mechanism that's calledCSMA-CD for carrier sense multiple access with collision detection,in which every station transmits and then senses the cableto detect if there has been a collision with other transmitter.It initially competed with other two technologies,called token ring and token busto be the standard of a local area networks.So all three were included in IEEE 802 project tocreate a standard for local area networks and ethernet was assignedthe 802.3 code.And became a standard in nineteen eighty five.But ethernet was able to adapt to market realities and shift toinexpensive and ubiquitous twisted pair wiring, so by the end of theeighties it was the dominant network technology.And today the vast majority of wired local networks use ethernetand even its fiber versions are being used by local telecommunicationcarriers for backbone transmissions.It's replacing other technologies and they are using ethernet for some backbonetransmissions. Ethernet specifications go up to layer twoof the OSI model so they specify the transmission media andthe physical and data link layer protocols.They have versions for coaxial, twisted pair and fiber optic cablingand these specifications set an universal 48 bitmac address and a standard frame format that has influencedother network standards. The first Ethernet specificationwas basedin using a shared coaxial cable to transmit at10 Mbits per second, detecting collisionsbetween transmitting stations so only one station could transmitat a time. This type of transmission is called half duplexor semi duplex, as each station can transmitand receive but not at the same time, only one can transmit,only one station is transmitting every time.This need to detect collisionsand react on time, because you have to stop transmitting if you detecteda collision, imposed a maximum distance between the network stations.And there were two standards for ethernet over coaxial cableten base five, that was also called thick ethernetbecause it used a thick coaxial cable, this was the first standardand then was later used for backbones because it reachedlonger distances. And the other one was called ten based twoalso called thin ethernet or cheapernetas it used a thinner coaxial cable and was much cheaper.With the development of structured cabling system for commercialpremises based in using individual shielded or unshieldedtwisted pairs for each station, ethernet evolved into a new standardcalled ten base-t, that was created to be able to use these new infrastructure.At first the network adapter cards for personal computers hadboth coaxial and twisted pair connectors, but soon the twisted pairtook over and coaxial cables where a thing of the past.The standard cabling has four pairs and at first only oneof them was used in half duplex mode,that is using a shared medium as with coaxial cables,for that they used devices called hubs that were able togive access to individual stations to this shared mediumusing a twisted pair, but soon two pairs where used and full-duplex,that means simultaneous transmission and reception was introducedtogether with devices called switches,that were able to give ten megabits per second to each connecteddevice instead of sharing the ten megabits per second among all.A standard for one hundred megabits per second transmissionover twisted pair called fast-ethernet orone hundred based-TX was introduced in nineteen ninety five and thenin nineteen ninety nine another standard called gigabit ethernetfor one thousand megabits per second, also name one thousand base-t followed.For older, lower quality cables this standards whatthey do is reach shorter distances with this speeds.Today gigabit ethernet adapters and switches arethe standard network equipment for desktop access,because they are cheap and they are reliable. And in two thousandsix a ten G base-t standard, that's aten gigabits per second standard was introduced,to have this speed over high quality twisted pairs.Now we are going to speak about ethernet over fiber-opticsthe one hundred megabits per second fast ethernet standard had alreadyseveral versions for different types of fiber optic cablingfor example, one hundred base fx, one hundred basesx these two for multi-mode fibers or one hundred base LX10for single-mode fibers. Gigabit ethernet also hasfiber versions for multi-mode fibersone thousand base fx and one thousand base sxand for single-mode fibers, one thousandbase LX10 for example. So depending on the type of fiberand wavelength we're going to use, we can use one standard withcheaper electronics or one standard with more expensiveelectronics but that can reach longer distances.Fiber optics gigabit ethernet are usuallyused for backbone communications inside corporate networks,because the connectors and electronics are more expensive thanwith twisted pair. And in two thousand two several 10Gbit persecond ethernet standard were launched and today there are standardsfor lan and for wan, that's it for local area networks and widearea networks, because they have been so successful that theyare replacing other access technologies in wannetworks of telecommunication operators.

 OSI Stack







Let's look at OSI stack from computer networks.So switching our smartphone and sending an email to a friend that lives aboutfive hundred kilometers from our homeis actually something we take for grantedand we just do it and it seems easy but it involves a lot of very complex processesand has to take into account a myriad of different situationsand so, as in any engineering problem, the best solution to tacklesuch complex thing is to break it down in a series of small problems that areeasier to solve and assemble the solution later.What we do to do this is to make an abstractionand separate different functions of a computer communicationprocess in layers. Considering that a layer communicates with the correspondinglayer and the other layer by using the services of the layer thatlies immediately under it, without knowing anything of whatlies below that and how it works. That way a programmer of software that managesthat layer only has to deal with how to communicate with itspartner layer in the other computer and how to use the servicesof the layer below, being completely isolated from the restof the processes involved in the communication and hencethe difficulties.So way to do this is the OSI,the Open System Interconnection model.Which has divided up the problem of communication into sevenabstraction layers. It defines a networking framework to implementprotocols in seven layers and control is passed from one layerto the next starting at the application layer in one stationand proceeding to the bottom layer over the channel to the next stationand back up the hierarchy. The OSI model doesn't really perform anyfunctions into the networking process, it's a conceptual frameworkthat allows the complex problem of computer internet to be brokendown into smaller more manageable problems.So this OSI model you see here on the slide, tales the taskof internetworking and divides that up into what is referred to as a verticalstack, that consists of seven layers that you can see here.So the bottom layer is the physical layer, on top of that thedata link layer, we have the network layer, the transport layer,the session layer, the presentation layer, and finally the applicationlayer. So these are again divided up basically in two sets andwe're going to look at each one of them.The first set is the application set and consists of the tree top layers ofthe model and this set deals with communication between applicationsin the two end-computers and it's three layers manage data not yetdivided in segments or packet. So the top layer is the applicationlayer so the application layer supports application and user processes.Communication partners are identified, quality of services is agreed uponuser authentication and privacy are considered,and any constraints on data syntax are identified.Everything at this layer is application-specificand it provides applications services for file transfer, emailand other networks software services.So telnet and ftp are applications that, for example, existentirely in the applicationlayer. So below that layer is the presentation layer. This layerprovides independence from differences in data representationlike for example encryption and it does this by translating fromapplication to networking format and vice versa.So the presentation layer works to transform data into the form thatthe application can accept it. This layer formats and encrypts datato be send across the network and provides freedom from compatibilityproblems. So it's sometimes called even the syntax layer becauseof that. The last layer in this application set is the sessionlayer. This layer establishes and managesand terminate connections between applications.So the session layer sets up, coordinates and terminate conversations,exchanges and dialogues between applications at eachend of the communication channel.It also deals with session and connection coordination between oneapplication and its counterpart in other computer.So the next set of layers are the remaining four layers which is calledthe transport set and it consists of the remaining layersof the model and deals with all the processes that we need when communicatingin a computer network. So these layers work which smaller chunksof information that are created to facilitate communication for internetworkingas we explain before, was needed. So the upper layer of this setis the transport layer. This layer providestransparent transfer of data between end systems or hosts and is responsible for end to end, error recovery and flow control.And this is actually the lower layer that works between end computersand it ensures complete data transfer.So below transport layer is the network layer and this layer offersits services to the transport layer, which is on top of it.It provides the switching and routing technologies and it providesmanaging multi-node networks, it creates logical paths, virtualcircuits and it can be used for transmitting data from nodeto node. So routing and forwarding are functions of thislayer as well as addressing, internetworking,error handling, congestion control and packet sequencingand this layer does not only work in end computers, but it alsoworks with intermediate network devices.So approaching the communications hardware, going downin the OSI model we find the data link layer and the data link layeris in charge of direct communication betweentwo adjacent nodes. At this layerdata packets are encoded and decoded into bits.It furnishes transmission protocol knowledge and manages and handleserrors in the physical layer, which is below.It does flow control and also frame synchronization.So the data link layer is divided again in two sub-layers,one is the media access control layer which is the mac layerand the logical link control layer the LLC layer.The MAC sublayer controls how a computer on the network gains accessto data and permissions to transmit it.The LLC layer controls frame synchronization, flow controland error checking. So finally, the layer that controls communicationwith a specific hardware is called the physical layer. Thislayer conveys the bitstream, which is anelectric impulse, light or radio signal, through the network atthe electric and mechanical level.So this layer provides the hardware means of sending and receivingdata on a carrier, including defining cables, cards and physicalaspects. So Gigabit Ethernet, RS232,DSL or ATM are protocols that you can specificallyfind at these physical layer component.So, to look at an example of how these different layers are implementedin different devices, here we see on the slide for example arouter that is able to connect different networks. So the router that is ableto connect different networks only goes up to the network layer,because it doesn't need to go beyond because the application layeris at the host. A switch that works inside the local networkonly goes up to the second layer, and the host applications obviouslyon this side and on the other side they have to go up to theapplication levels and to the seven layers.So with the widespread use of the internet the simpler tcp/ipprotocols stack has become the default in standard internetworking acrossthe world and so OSI model is no longer used to develop communicationsoftware but it's still a very useful tool in educationsince the abstraction layers make it simpler to differentiatehow network communication works. So we can see it's used whensystem administrators troubleshoot aa network problem, for example. So first of all they say is it a physicalproblem, is the network cable plugged in.Then they go on and say is that a datalink problemdo you have a link light in the network cardThe third thing they can look at is if it is a network problem.Is the pc getting an ip address Then the fourth thing iswith the transport. Can you ping your default gateway Would be a specificquestion at that level. And then we can look at the session and we canask questions like well do you have a dns server information.Can you ping 4.2.2.2 but not google.comAnd finally if this all doesn't work they can ask questions inthe presentation application layer like for example can you reacha website with your web browser. So the OSI models stack is very usefulin trying to understand how it all works because of these sevenabstraction layers. So we have seen the OSI stack

TCP/IP Stack

Hi, we are going to speak about the TCP/IP stack in this unit.TCP/IP stack is the family of protocols that rule the current internetwhile other protocols are also used in computer networks,TCP/IP is by far the most common of them.TCP/IP can be compared to the OSI model as a simpler and more efficientimplementation of the conceptual seven-layer models of OSI.TCP/IP deals instead with seven with four layers of communications,anf data packets are passed top down for transmission as in OSI.TCP/IP doesn't specify the network access layer so it's quitecommon to mix the OSI model and the TCP/IP model and talk aboutphysical layer and data link layer instead of a network accesslayer. Here you can see the comparison of the twostacks.Here we can seean example of the data flow in remote shellsession logging, someone that is accesing a remotecomputer to control it from a shell sessionusing TCP/IP. Each layer as you can see here addsheaders and creates a new data structure encapsulating the one thatwas handled by the layer that's over it.We can see how data is encapsulated for example here inthe second figure. We can see how the data fromapplication-layer is encapsulated in transport layer witha header then data from trasport layer is encapsulated in internet layerwith another header and then it is encapsulated in the data link layerwith another header and footer in this case.On reception we can see here, each layer strips the header added by its counterpartin the sending device, then performsthe operations that are assigned to it and thenpasses the resulting data to thelayer over it. Look you can see it here, itdecodes it and then pass the resulting data structure to the layerover it. Data structures have different names in differentcommunication layers, for example here is calledframe, here it's called datagram, here is called segment.If the connection has to pass through severalnetwork devices they use the same mechanism but onlyup to the level in which they work for example a level three device,a router, that is a device that is able to connect todifferent networks, it only goes up to internet layer,that is called  network layer in OSI modelas you can see here, and then uses the othernetwork under it to communicate with the next device.Here we don`t have any, but if the connection was made at leveltwo in the OSI model, that is something that switches do,they work inside the local network,so they only go up to OSI layer two to communicate with theother one. This is inside the network access layer in the TCP/IPmodel. The hosts as you can see, useall layers to communicate from one process to another process.Well, how are standards made in InternetStandards are made by IETF created in 1992,that uses a method that's called requests for commentsRFC, that is a consensual standard-setting processwhere standards are created in consensus with a broadgroup of different entities. And then are named withRFC and a four digits code that nowadays is in theseven thousands for the last standards.The RFC-0001  dates from April nineteen sixty nine.All this TCP/IP protocols used in Internet are defined inthe public RFC that are constantly updated.Well, we are going to speak about internet addresses now.This is something that we also saw in the first course of theISC series, but we're gonna and revisit it because they are partof the TCP/IP protocol. Internet addresses are use to locate thedestination computer in order to send packetsthrough the Internet, in some sense is like the postal address ofcomputer. And every device on the Internet has assigned an IP address.In the late sixties, when internet startedip addresses with four bytes, that is with32 bits, as a byte has eight bits,were used, because the number of possible different addresses, thatis two to the power of 32, that is4.3 US billion, was considereda huge number that was more than enoughfor all the computers that were foreseen.But with the rapid growth of the internet after commercializationin the nineties, it became evident that the address space availablefor IPv4  was not enough to connectall the new devices in the future. And by nineteen ninety eightthe IETF, internet engineering task force,formalized a successor protocol that was called IPV6that instead of 32 bits uses128 bits so it allows fortwo to the power of 128,that is more than 7.9by ten to the power of 28 timesaddresses as IPV4 had. This more than enoughfor now and for the future. And now we have two differenttypes of IP addresses the IPV4that were the first ones and the IPV6, the last ones.The plan was that today allthe IP addresses were going to be IPV6 because they were planningsomething like two thousand and twelveto reach the point at all addresseswould have change. But it hasn't been like that because IPV4 addresseshave been used for more devices that when theythought they could, using very inteligent mechanismsand the adoption of the IPV6 has beenslower than was thought. Nowadays perhaps something like, I think it is fifteen percent of theIP traffic that arrives to google serversit's IPV6. Both protocols are notinteroperable but they have devised gateways that can communicatemachines using IPV6 with machines using IPV4 and mostmodern operating systems, for example linux, windowsand iOS, are able to support bothIPV6 and IPV4.Over the internet layer we find the transportlayer. Transport layer is the first that is between hosts.From end host to end host. In TCP/IP transport layer we find two differentprotocols, one is tcp fortransmission control protocol that this connection-orientedwhat means that it transmits data as a byte stream,providing flow control. It guarantees thatthe data transferred remains intact and that it arrivesin the same order in which it was send. The other protocol that iscalled UDP for user datagram protocolit's a connectionless protocol that transmits the packetsindividually and gives no guarantees at all, but is faster and lightweight.So, for some uses it's better TCP and for other uses is better UTP.The transport layer addresses allow users to connect to an specificapplication in a computers host, and they are known in TCP/IP as socketsor ports. A socket is a combination of an IP address,,the IP address of the host, with a number that identifies the specificapplication or service. This is separated, as you can see in the image, bya colon. Specific port numbers are often used to identifyspecific services. Of the thousands of enumerated portsthere are one thousand and twenty four well known port numbersthat are reserved by convention to identify specific serviceson hosts, for example for telnet, for ftpfor smtp, that is the mail, or for web protocol that is http.We don't usually see these port numbers because every applicationis programmed to use a specific port, so when we use an ip addressin a web browser to access a websiteit understands that the requests will go to port 80 by default.But we could have for example two differentweb services in the same machine with the same IP address and we couldbe using a port, for example, eighty eighty for the secondwebserver and we can put that in the address bar of a web browser.Well, TCP/IP protocol suite encloses far more protocolsthan the basic ones we have commented. You can see herein the image some protocols that use TCP, other protocols thatuse UDP for example, you can see DNS, FTP,DHCP... in the other side you can see for exampleHTTP and here in the image you can see how does this fitin OSI seven-layer model.

UPValenciaX: ISC101.2x Information Systems and Computer Applications, Part 2: Hardware