1 1 00:00:00,280 --> 00:00:02,550 Layer 4, the transport layer. 2 2 00:00:02,550 --> 00:00:05,120 Now, the transport layer is our dividing line 3 3 00:00:05,120 --> 00:00:08,110 between what we call the upper layers of the OSI model 4 4 00:00:08,110 --> 00:00:10,630 and the lower layers of that OSI model. 5 5 00:00:10,630 --> 00:00:12,350 Now, we've already covered the lower layers 6 6 00:00:12,350 --> 00:00:13,780 when we talked about the physical, 7 7 00:00:13,780 --> 00:00:15,790 the data link, and the network layers. 8 8 00:00:15,790 --> 00:00:18,370 And so now, we're going to move into the upper layers, 9 9 00:00:18,370 --> 00:00:20,910 starting with this layer, the transport layer, 10 10 00:00:20,910 --> 00:00:23,200 the session layer, the presentation layer, 11 11 00:00:23,200 --> 00:00:24,920 and the application layer. 12 12 00:00:24,920 --> 00:00:26,500 Now, in the next couple of lessons, 13 13 00:00:26,500 --> 00:00:28,789 we're going to cover each of these as we go forward. 14 14 00:00:28,789 --> 00:00:31,210 Now, segment is our datatype here 15 15 00:00:31,210 --> 00:00:33,400 when we're dealing with the transport layer. 16 16 00:00:33,400 --> 00:00:35,490 When we deal with segments and datagrams, 17 17 00:00:35,490 --> 00:00:38,010 we're talking about the transport layer. 18 18 00:00:38,010 --> 00:00:39,560 Now, as we talk about datagrams, 19 19 00:00:39,560 --> 00:00:42,350 we're going to go into those a little bit more later in depth. 20 20 00:00:42,350 --> 00:00:44,710 But for now, let's focus on the two protocols 21 21 00:00:44,710 --> 00:00:46,750 that we have inside layer 4, 22 22 00:00:46,750 --> 00:00:49,820 which are the TCP and the UDP protocols. 23 23 00:00:49,820 --> 00:00:50,980 And we're also going to introduce 24 24 00:00:50,980 --> 00:00:53,440 a couple of extra reliability features here 25 25 00:00:53,440 --> 00:00:55,910 known as windowing and buffering. 26 26 00:00:55,910 --> 00:00:58,180 Now, what is TCP? 27 27 00:00:58,180 --> 00:01:01,160 TCP is a transmission control protocol. 28 28 00:01:01,160 --> 00:01:03,240 It is a connection-oriented protocol, 29 29 00:01:03,240 --> 00:01:06,330 which means it's a reliable way to transport segments 30 30 00:01:06,330 --> 00:01:07,980 across our network. 31 31 00:01:07,980 --> 00:01:09,490 Now, if a segment has dropped, 32 32 00:01:09,490 --> 00:01:11,670 the protocol will actually ask for acknowledgment 33 33 00:01:11,670 --> 00:01:12,930 each and every time. 34 34 00:01:12,930 --> 00:01:14,780 If it doesn't get that acknowledgement, 35 35 00:01:14,780 --> 00:01:17,430 it's going to resend that piece of information. 36 36 00:01:17,430 --> 00:01:20,320 That's why we call this connection full protocol 37 37 00:01:20,320 --> 00:01:22,830 because it has this two-way type of information 38 38 00:01:22,830 --> 00:01:24,170 where I'm sending you information 39 39 00:01:24,170 --> 00:01:26,080 and I'm verifying you actually got it 40 40 00:01:26,080 --> 00:01:29,070 by listening that you got it and you give me a response. 41 41 00:01:29,070 --> 00:01:30,710 Now, let's look at this little diagram here 42 42 00:01:30,710 --> 00:01:31,795 on the screen for a second. 43 43 00:01:31,795 --> 00:01:34,120 You're going to see that I have a client on the left 44 44 00:01:34,120 --> 00:01:35,690 and a server on the right. 45 45 00:01:35,690 --> 00:01:38,750 Now, the client is going to send what's called a SYN packet 46 46 00:01:38,750 --> 00:01:41,537 or a synchronization packet over to the server. 47 47 00:01:41,537 --> 00:01:43,120 Now, when the server gets that, 48 48 00:01:43,120 --> 00:01:45,810 it's going to send back a synchronization acknowledgment 49 49 00:01:45,810 --> 00:01:48,490 to the client known as a SYN-ACK. 50 50 00:01:48,490 --> 00:01:50,610 Now, when the client gets that acknowledgement, 51 51 00:01:50,610 --> 00:01:53,290 it's going to send back its own acknowledgement to the server. 52 52 00:01:53,290 --> 00:01:54,950 This is known as the ACK. 53 53 00:01:54,950 --> 00:01:57,830 Now, when we do this SYN, SYN-ACK, ACK, 54 54 00:01:57,830 --> 00:02:00,940 this is what we refer to as a three-way handshake. 55 55 00:02:00,940 --> 00:02:03,620 Essentially, it's the client going, hey server, 56 56 00:02:03,620 --> 00:02:05,310 are you ready to get some information? 57 57 00:02:05,310 --> 00:02:07,390 And then the server says, sure, why not? 58 58 00:02:07,390 --> 00:02:08,780 Send me some information. 59 59 00:02:08,780 --> 00:02:10,987 And the client says, okay, here it comes. 60 60 00:02:10,987 --> 00:02:13,540 And then, the transmission is going to begin 61 61 00:02:13,540 --> 00:02:15,710 because we've established that three-way handshake 62 62 00:02:15,710 --> 00:02:18,141 and we know that both sides are ready to communicate. 63 63 00:02:18,141 --> 00:02:19,900 Now, are you ready? 64 64 00:02:19,900 --> 00:02:22,422 Yes, I am, here it comes. 65 65 00:02:22,422 --> 00:02:25,460 Now, every time this data, which we call a segment, 66 66 00:02:25,460 --> 00:02:27,050 is sent across the network, 67 67 00:02:27,050 --> 00:02:29,360 there is going to be an acknowledgment that it was received 68 68 00:02:29,360 --> 00:02:31,380 and that tells us there was successful 69 69 00:02:31,380 --> 00:02:33,330 two-way communication occurring. 70 70 00:02:33,330 --> 00:02:35,760 Now, if the server is expecting to get 100 pieces 71 71 00:02:35,760 --> 00:02:38,800 of information, but it only got 98 of those, 72 72 00:02:38,800 --> 00:02:40,020 it's going to say to the client, 73 73 00:02:40,020 --> 00:02:42,448 hey, you told me you're going to send me 100 things 74 74 00:02:42,448 --> 00:02:44,570 but you only sent me 98. 75 75 00:02:44,570 --> 00:02:46,770 Send me over those two things that I'm missing. 76 76 00:02:46,770 --> 00:02:48,739 And then a retransmission occurs. 77 77 00:02:48,739 --> 00:02:51,191 This way, the communication can go forth 78 78 00:02:51,191 --> 00:02:53,000 and we can always make sure we're getting 79 79 00:02:53,000 --> 00:02:55,030 what we're supposed to because we have this resending 80 80 00:02:55,030 --> 00:02:56,881 of the packets across the network. 81 81 00:02:56,881 --> 00:02:59,500 Now, this is used for all network data 82 82 00:02:59,500 --> 00:03:03,150 that needs to be assured to get to its final destination. 83 83 00:03:03,150 --> 00:03:05,780 I like to think about this like certified mail. 84 84 00:03:05,780 --> 00:03:08,780 If I want to send a message to the IRS, for example, 85 85 00:03:08,780 --> 00:03:10,330 I want to make sure that they get it 86 86 00:03:10,330 --> 00:03:11,940 and that doesn't get lost in the mail. 87 87 00:03:11,940 --> 00:03:13,960 So, I might pay a little extra money 88 88 00:03:13,960 --> 00:03:16,530 to get a certified receipt that when it gets there, 89 89 00:03:16,530 --> 00:03:19,140 they have to sign it and that gets mailed back to me. 90 90 00:03:19,140 --> 00:03:21,080 This way, when I get that receipt back, 91 91 00:03:21,080 --> 00:03:23,381 I know that the IRS got my mail package. 92 92 00:03:23,381 --> 00:03:25,835 That's the way TCP works. 93 93 00:03:25,835 --> 00:03:28,970 Now, on the other hand, we have another protocol 94 94 00:03:28,970 --> 00:03:30,272 known as UDP. 95 95 00:03:30,272 --> 00:03:33,560 UDP is what we call a connectionless protocol, 96 96 00:03:33,560 --> 00:03:35,757 meaning, it doesn't have to wait for connections. 97 97 00:03:35,757 --> 00:03:39,150 UDP stands for User Datagram Protocol 98 98 00:03:39,150 --> 00:03:40,610 and the reason why we call it a datagram 99 99 00:03:40,610 --> 00:03:42,470 is because if you're using UDP, 100 100 00:03:42,470 --> 00:03:44,080 you're using this type of data, 101 101 00:03:44,080 --> 00:03:45,400 it's called a datagram. 102 102 00:03:45,400 --> 00:03:47,820 And so, for the exam, I want you to remember 103 103 00:03:47,820 --> 00:03:51,650 that layer 4 is for segments almost exclusively 104 104 00:03:51,650 --> 00:03:53,950 because we use it with TCP. 105 105 00:03:53,950 --> 00:03:58,040 But if you're using UDP, this is now called a datagram. 106 106 00:03:58,040 --> 00:04:00,500 So, if you have a datagram or a segment, 107 107 00:04:00,500 --> 00:04:01,681 you're in layer 4. 108 108 00:04:01,681 --> 00:04:05,940 Now, when we talk about UDP, UDP is unreliable 109 109 00:04:05,940 --> 00:04:08,420 and it transmits segments called datagrams 110 110 00:04:08,420 --> 00:04:09,430 and if they're dropped, 111 111 00:04:09,430 --> 00:04:11,950 the sender will never even know that it happened. 112 112 00:04:11,950 --> 00:04:13,760 Now, why would I want to send stuff 113 113 00:04:13,760 --> 00:04:15,220 where the sender isn't aware of it 114 114 00:04:15,220 --> 00:04:16,629 and I don't get any kind of receipt? 115 115 00:04:16,629 --> 00:04:20,800 Well, UDP is really good for audio and visual streaming 116 116 00:04:20,800 --> 00:04:22,830 because you send a lot of data 117 117 00:04:22,830 --> 00:04:25,530 and there's a lot less overhead when you use UDP 118 118 00:04:25,530 --> 00:04:27,770 because we don't have that constant three-way handshake 119 119 00:04:27,770 --> 00:04:29,420 to establish it and we don't have 120 120 00:04:29,420 --> 00:04:31,900 all the checks and balances that are associated 121 121 00:04:31,900 --> 00:04:33,431 by using TCP. 122 122 00:04:33,431 --> 00:04:37,070 So, by using UDP, you can really increase the performance 123 123 00:04:37,070 --> 00:04:39,140 of your network because you're going to have 124 124 00:04:39,140 --> 00:04:40,720 zero retransmissions, 125 125 00:04:40,720 --> 00:04:42,830 you're just going to end up dropping information. 126 126 00:04:42,830 --> 00:04:44,410 Now, isn't that a bad thing? 127 127 00:04:44,410 --> 00:04:46,470 Why would we want to drop information? 128 128 00:04:46,470 --> 00:04:49,670 Well, for certain applications, it really doesn't matter. 129 129 00:04:49,670 --> 00:04:52,240 For example, you're streaming this video right now 130 130 00:04:52,240 --> 00:04:55,050 and if I dropped out for one hundredth of a second, 131 131 00:04:55,050 --> 00:04:56,470 would you even notice? 132 132 00:04:56,470 --> 00:04:57,850 Well, you probably wouldn't. 133 133 00:04:57,850 --> 00:04:59,720 And that's why UDP is so good 134 134 00:04:59,720 --> 00:05:02,450 because we can drop one 100th of the time here 135 135 00:05:02,450 --> 00:05:04,200 and you're really never even going to notice it 136 136 00:05:04,200 --> 00:05:06,050 and there won't be a retransmission. 137 137 00:05:06,050 --> 00:05:09,240 But with TCP, it's going to lead to a lot more buffering 138 138 00:05:09,240 --> 00:05:11,590 because you have to wait and then get resent to you 139 139 00:05:11,590 --> 00:05:14,170 and then put it in the right place and then play it back. 140 140 00:05:14,170 --> 00:05:16,700 And so, because of that acknowledgement and that overhead 141 141 00:05:16,700 --> 00:05:18,810 for every single second of this video, 142 142 00:05:18,810 --> 00:05:20,840 it's going to end up making a lot larger 143 143 00:05:20,840 --> 00:05:22,700 and use a lot more bandwidth. 144 144 00:05:22,700 --> 00:05:25,180 And that's one of the big reasons why we use UDP 145 145 00:05:25,180 --> 00:05:27,590 for video streaming and audio streaming. 146 146 00:05:27,590 --> 00:05:31,680 Now, let's do a quick little summary here of TCP versus UDP 147 147 00:05:31,680 --> 00:05:34,320 because this is a really, really important concept. 148 148 00:05:34,320 --> 00:05:36,550 In fact, if you have your notes out right now, 149 149 00:05:36,550 --> 00:05:38,080 I would write down this chart 150 150 00:05:38,080 --> 00:05:39,360 that I'm going to tell you right now 151 151 00:05:39,360 --> 00:05:41,470 as we talk about TCP versus UDP 152 152 00:05:41,470 --> 00:05:43,710 because it really is that important. 153 153 00:05:43,710 --> 00:05:46,300 Now first, TCP is reliable. 154 154 00:05:46,300 --> 00:05:49,970 It has a three-way handshake where UDP is not very reliable, 155 155 00:05:49,970 --> 00:05:51,470 it's an unreliable protocol 156 156 00:05:51,470 --> 00:05:54,090 because there is no three-way handshake. 157 157 00:05:54,090 --> 00:05:56,440 TCP is what we call connection-oriented 158 158 00:05:56,440 --> 00:05:58,154 or a connectionfull protocol 159 159 00:05:58,154 --> 00:05:59,810 because we have that three-way handshake 160 160 00:05:59,810 --> 00:06:03,570 in the acknowledgements but UDP is connectionless. 161 161 00:06:03,570 --> 00:06:05,490 It's a fire and forget method. 162 162 00:06:05,490 --> 00:06:07,080 I just start sending out information 163 163 00:06:07,080 --> 00:06:09,204 and hopefully you're going to get it. 164 164 00:06:09,204 --> 00:06:13,120 TCP uses segment retransmission and flow control 165 165 00:06:13,120 --> 00:06:14,670 that's being handled through windowing, 166 166 00:06:14,670 --> 00:06:16,710 which we're going to talk about more in just a second. 167 167 00:06:16,710 --> 00:06:19,990 UDP, on the other hand, there is no retransmission 168 168 00:06:19,990 --> 00:06:21,470 and no windowing. 169 169 00:06:21,470 --> 00:06:24,380 With TCP, we have segmentation of our sequencing 170 170 00:06:24,380 --> 00:06:26,050 of all of our different segments. 171 171 00:06:26,050 --> 00:06:28,620 With UDP, there is no sequencing. 172 172 00:06:28,620 --> 00:06:30,770 Now, what this means is, as I send everything out, 173 173 00:06:30,770 --> 00:06:33,980 I'm going to send it out in the proper order from one to 100. 174 174 00:06:33,980 --> 00:06:36,068 I'll do this for both TCP and UDP. 175 175 00:06:36,068 --> 00:06:38,150 Now, if you miss some of those pieces, 176 176 00:06:38,150 --> 00:06:39,600 or they arrive in a different order 177 177 00:06:39,600 --> 00:06:41,720 because they take different paths over the network, 178 178 00:06:41,720 --> 00:06:43,740 with TCP, they're sequencing 179 179 00:06:43,740 --> 00:06:45,387 so it knows that you have one to 1,000 180 180 00:06:45,387 --> 00:06:47,640 and it puts them back in the right sequence. 181 181 00:06:47,640 --> 00:06:50,090 With UDP, whatever they come in as, 182 182 00:06:50,090 --> 00:06:51,830 that's how it's going to broadcast it. 183 183 00:06:51,830 --> 00:06:55,752 And so, it can be coming in 1, 50, 2, 500, 184 184 00:06:55,752 --> 00:06:59,050 3, 4, 5, 6, 20, 185 185 00:06:59,050 --> 00:07:00,450 in any random order like that 186 186 00:07:00,450 --> 00:07:01,820 and that's how you're going to hear it. 187 187 00:07:01,820 --> 00:07:04,260 So, if video, you may hear a little bit of jumpiness, 188 188 00:07:04,260 --> 00:07:06,010 or a little bit of high-pitch squeaks 189 189 00:07:06,010 --> 00:07:06,960 or something like that 190 190 00:07:06,960 --> 00:07:09,055 because one of those frames may have come out of order. 191 191 00:07:09,055 --> 00:07:11,930 Now, when we go back to TCP, it is going to acknowledge 192 192 00:07:11,930 --> 00:07:13,210 each of those segments. 193 193 00:07:13,210 --> 00:07:15,000 And so, we have acknowledgment. 194 194 00:07:15,000 --> 00:07:17,040 If I don't get it, I know that I didn't get it 195 195 00:07:17,040 --> 00:07:20,230 and I can get it retransmitted to me and then get it again. 196 196 00:07:20,230 --> 00:07:22,440 With UDP, there is no acknowledgement. 197 197 00:07:22,440 --> 00:07:25,410 So, again, UDP has a lot less overhead 198 198 00:07:25,410 --> 00:07:26,790 because there's no connection, 199 199 00:07:26,790 --> 00:07:30,000 no windowing, no retransmission, no sequencing, 200 200 00:07:30,000 --> 00:07:31,332 and no acknowledgement. 201 201 00:07:31,332 --> 00:07:33,850 Now, if you have to get something there 202 202 00:07:33,850 --> 00:07:36,110 and you want to make sure the person got it, 203 203 00:07:36,110 --> 00:07:39,356 you really have to use TCP as your protocol of choice. 204 204 00:07:39,356 --> 00:07:42,160 And that's why we really are going to use TCP 205 205 00:07:42,160 --> 00:07:44,760 for things like banking and websites and ecommerce 206 206 00:07:44,760 --> 00:07:45,990 and things like that. 207 207 00:07:45,990 --> 00:07:48,130 But if we have something that has a lot of data 208 208 00:07:48,130 --> 00:07:49,760 like audio or video streaming, 209 209 00:07:49,760 --> 00:07:51,780 UDP really does well with that 210 210 00:07:51,780 --> 00:07:52,760 because we don't need to get 211 211 00:07:52,760 --> 00:07:54,550 every single piece of that file. 212 212 00:07:54,550 --> 00:07:57,760 We can skip a little bit here and there and that's okay. 213 213 00:07:57,760 --> 00:07:58,910 Now, earlier in the lesson, 214 214 00:07:58,910 --> 00:08:01,070 I mentioned a concept known as windowing 215 215 00:08:01,070 --> 00:08:02,730 and I said, we'd get to it later. 216 216 00:08:02,730 --> 00:08:04,040 Well, here we are. 217 217 00:08:04,040 --> 00:08:05,670 We're going to talk about windowing. 218 218 00:08:05,670 --> 00:08:07,270 Now, what is windowing? 219 219 00:08:07,270 --> 00:08:08,770 Windowing is going to allow the clients 220 220 00:08:08,770 --> 00:08:11,270 to adjust the amount of data in each segment 221 221 00:08:11,270 --> 00:08:13,051 as it goes through the transmission. 222 222 00:08:13,051 --> 00:08:15,120 This way, we can continually adjust 223 223 00:08:15,120 --> 00:08:17,650 to either send more or fewer pieces of data 224 224 00:08:17,650 --> 00:08:19,813 for each segment that's being transmitted. 225 225 00:08:19,813 --> 00:08:21,820 So, the whole idea here with windowing 226 226 00:08:21,820 --> 00:08:23,410 is that if you're sending data 227 227 00:08:23,410 --> 00:08:25,001 and you're getting a lot of retransmissions, 228 228 00:08:25,001 --> 00:08:27,560 well, you might be sending too much information. 229 229 00:08:27,560 --> 00:08:28,930 So, you need to back that down 230 230 00:08:28,930 --> 00:08:30,630 and close the window a little bit. 231 231 00:08:30,630 --> 00:08:32,420 So, you'll send less each time. 232 232 00:08:32,420 --> 00:08:34,410 Now, if you're not getting any retransmissions, 233 233 00:08:34,410 --> 00:08:36,390 it means you're probably not going fast enough. 234 234 00:08:36,390 --> 00:08:38,230 So, instead, we can open up that window 235 235 00:08:38,230 --> 00:08:40,737 and send more data with each of those segments. 236 236 00:08:40,737 --> 00:08:41,910 And then, if we start getting 237 237 00:08:41,910 --> 00:08:44,010 a lot more of those retransmissions happening again, 238 238 00:08:44,010 --> 00:08:46,410 well, we start closing that window down. 239 239 00:08:46,410 --> 00:08:48,306 And so, always we're opening and closing the window 240 240 00:08:48,306 --> 00:08:51,900 to maximize our throughput and our bandwidth here. 241 241 00:08:51,900 --> 00:08:54,020 So, if you ever copied a file over a network 242 242 00:08:54,020 --> 00:08:55,890 on a Windows machine, you've probably seen 243 243 00:08:55,890 --> 00:08:58,530 where it starts that movie file and it starts saying, 244 244 00:08:58,530 --> 00:09:00,350 hey, you have 20 minutes remaining 245 245 00:09:00,350 --> 00:09:02,460 and then it drops down to five minutes remaining, 246 246 00:09:02,460 --> 00:09:04,230 then it jumps up to 50 minutes remaining, 247 247 00:09:04,230 --> 00:09:06,750 and then 30 minutes remaining, and then an hour, 248 248 00:09:06,750 --> 00:09:07,972 and then it goes down to three minutes. 249 249 00:09:07,972 --> 00:09:10,170 And it has a really hard time estimating 250 250 00:09:10,170 --> 00:09:12,990 how long it's going to take to move that really large file 251 251 00:09:12,990 --> 00:09:15,640 off of your shared drive and onto your Windows computer. 252 252 00:09:15,640 --> 00:09:16,900 Now, why is that? 253 253 00:09:16,900 --> 00:09:18,980 Well, that's windowing at work. 254 254 00:09:18,980 --> 00:09:21,020 What's happening here is that as there's issues 255 255 00:09:21,020 --> 00:09:23,830 on the network, and there's a lot more retransmissions, 256 256 00:09:23,830 --> 00:09:26,580 the window decreases become smaller 257 257 00:09:26,580 --> 00:09:28,960 and that means, we have to send more segments 258 258 00:09:28,960 --> 00:09:31,440 to get all that data across, which takes more time. 259 259 00:09:31,440 --> 00:09:33,660 Now, as things go better, 260 260 00:09:33,660 --> 00:09:35,390 and the network starts flowing again, 261 261 00:09:35,390 --> 00:09:37,940 that window opens up and we can send less segments 262 262 00:09:37,940 --> 00:09:40,470 with more data each time and that's going to end up 263 263 00:09:40,470 --> 00:09:42,624 decreasing the time or making it go down. 264 264 00:09:42,624 --> 00:09:46,160 So, what happens here is as you can see on the screen, 265 265 00:09:46,160 --> 00:09:48,450 let's say that little green thing is what I'm sending 266 266 00:09:48,450 --> 00:09:50,360 and I start sending the information over 267 267 00:09:50,360 --> 00:09:52,610 but that red starts creeping up 268 268 00:09:52,610 --> 00:09:55,040 to where we start to not being able to keep up with it. 269 269 00:09:55,040 --> 00:09:58,280 So, we'll come back down and then the red can creep up again 270 270 00:09:58,280 --> 00:10:00,592 and then, we'll come back down and we'll keep doing that. 271 271 00:10:00,592 --> 00:10:02,520 Hopefully, the red and the green here 272 272 00:10:02,520 --> 00:10:04,660 will eventually match at a higher level 273 273 00:10:04,660 --> 00:10:05,503 than it was starting with. 274 274 00:10:05,503 --> 00:10:08,680 As soon as we open up that window and close that window, 275 275 00:10:08,680 --> 00:10:11,190 we can start out slow and then we can go faster 276 276 00:10:11,190 --> 00:10:13,760 by opening up that window and faster and faster 277 277 00:10:13,760 --> 00:10:15,600 until we have problems and then we'll start 278 278 00:10:15,600 --> 00:10:16,800 closing it down again. 279 279 00:10:16,800 --> 00:10:19,550 And we'll keep doing that over and over and over again, 280 280 00:10:19,550 --> 00:10:21,660 until we get the best bandwidth we can 281 281 00:10:21,660 --> 00:10:24,180 as we try to push as much as we can. 282 282 00:10:24,180 --> 00:10:27,170 So, for example, if I started counting numbers to you, 283 283 00:10:27,170 --> 00:10:28,800 I'm going to start going slow. 284 284 00:10:28,800 --> 00:10:32,695 One, two, three, that's pretty slow, right? 285 285 00:10:32,695 --> 00:10:35,000 You'll say okay, okay, I got it, Jason. 286 286 00:10:35,000 --> 00:10:35,950 You can go faster. 287 287 00:10:35,950 --> 00:10:37,490 So, I start talking faster. 288 288 00:10:37,490 --> 00:10:41,220 One, two, three, four, five, okay? 289 289 00:10:41,220 --> 00:10:42,860 That's still good, let's try again. 290 290 00:10:42,860 --> 00:10:43,750 One, two, three, four, five, 291 291 00:10:43,750 --> 00:10:45,950 oh, wait, wait, wait, wait, that's too fast, Jason. 292 292 00:10:45,950 --> 00:10:47,440 Okay, let me slow down. 293 293 00:10:47,440 --> 00:10:50,230 One, two, three, four, five, you got it? 294 294 00:10:50,230 --> 00:10:51,960 Okay, and we keep doing that. 295 295 00:10:51,960 --> 00:10:53,250 That's the idea of windowing. 296 296 00:10:53,250 --> 00:10:54,750 I'll speed up and I'll slow down 297 297 00:10:54,750 --> 00:10:56,560 until you don't have any errors copying down 298 298 00:10:56,560 --> 00:10:57,560 what I'm saying. 299 299 00:10:57,560 --> 00:10:59,150 That's the whole idea here. 300 300 00:10:59,150 --> 00:11:01,360 And we want to be able to send as much information to you 301 301 00:11:01,360 --> 00:11:02,940 as quickly as possible 302 302 00:11:02,940 --> 00:11:05,000 with the least amount of retransmissions 303 303 00:11:05,000 --> 00:11:07,323 but still getting the maximum throughput. 304 304 00:11:07,323 --> 00:11:10,670 Now, the next concept we want to talk about is buffering. 305 305 00:11:10,670 --> 00:11:12,310 If you've ever watched online video, 306 306 00:11:12,310 --> 00:11:14,450 you've probably dealt with buffering before. 307 307 00:11:14,450 --> 00:11:17,980 Now, devices such as routers have a special memory in them 308 308 00:11:17,980 --> 00:11:19,980 that will store segments if the bandwidth 309 309 00:11:19,980 --> 00:11:21,540 isn't readily available. 310 310 00:11:21,540 --> 00:11:23,270 Now, this is called the buffer. 311 311 00:11:23,270 --> 00:11:24,750 So, when it becomes available, 312 312 00:11:24,750 --> 00:11:26,920 it'll go ahead and start transmitting out the contents 313 313 00:11:26,920 --> 00:11:29,240 of that buffer and clear itself out. 314 314 00:11:29,240 --> 00:11:31,500 The same thing happens when you try to load a video. 315 315 00:11:31,500 --> 00:11:32,954 If the network is congested, 316 316 00:11:32,954 --> 00:11:35,510 it will take in a lot of information at first 317 317 00:11:35,510 --> 00:11:37,130 in anticipation of the fact 318 318 00:11:37,130 --> 00:11:38,750 that you're going to watch it faster, 319 319 00:11:38,750 --> 00:11:41,390 then, you're going to be able to download the rest of the video. 320 320 00:11:41,390 --> 00:11:43,030 And so, this is the idea with our buffering 321 321 00:11:43,030 --> 00:11:44,450 our routers, as well. 322 322 00:11:44,450 --> 00:11:46,910 So, if the buffer is going to overflow, though, 323 323 00:11:46,910 --> 00:11:49,130 and on a router you only have so much space 324 324 00:11:49,130 --> 00:11:51,070 and you start putting too much information there 325 325 00:11:51,070 --> 00:11:52,410 cause you can't send it out, 326 326 00:11:52,410 --> 00:11:53,940 then what's going to end up happening? 327 327 00:11:53,940 --> 00:11:55,080 You run out of memory. 328 328 00:11:55,080 --> 00:11:57,603 And when you run out of memory, the segment's will drop. 329 329 00:11:57,603 --> 00:11:59,980 So, let's look at an example of how buffering 330 330 00:11:59,980 --> 00:12:01,290 is going to work. 331 331 00:12:01,290 --> 00:12:03,350 Let's take a look at the buffer on the router 332 332 00:12:03,350 --> 00:12:06,150 and here we have router four in our diagram. 333 333 00:12:06,150 --> 00:12:08,938 Notice how it's kind of the central point of this diagram. 334 334 00:12:08,938 --> 00:12:12,480 Now, I have stuff coming into it from router number six, 335 335 00:12:12,480 --> 00:12:15,050 and router number one and router number three. 336 336 00:12:15,050 --> 00:12:17,770 So, if I look at all of those, there's 100 megabits, 337 337 00:12:17,770 --> 00:12:20,230 100 megabits, and 10 megabits. 338 338 00:12:20,230 --> 00:12:23,720 That's a possibility of 210 megabits per second 339 339 00:12:23,720 --> 00:12:26,219 of information going into router four. 340 340 00:12:26,219 --> 00:12:28,580 Now, if it needs to send that information out 341 341 00:12:28,580 --> 00:12:32,230 to router five, and there's only a 50 megabit connection, 342 342 00:12:32,230 --> 00:12:33,510 you can see pretty quickly 343 343 00:12:33,510 --> 00:12:35,437 that there's going to be a bottleneck here. 344 344 00:12:35,437 --> 00:12:37,500 Now, what's going to end up happening for us 345 345 00:12:37,500 --> 00:12:39,530 is that router four is going to have to catch 346 346 00:12:39,530 --> 00:12:42,190 all that extra information in its buffer. 347 347 00:12:42,190 --> 00:12:43,810 And when it has more availability, 348 348 00:12:43,810 --> 00:12:46,060 it'll send that information out to router five 349 349 00:12:46,060 --> 00:12:47,294 and clear its buffer. 350 350 00:12:47,294 --> 00:12:51,410 You may ask, why would we design a network this way? 351 351 00:12:51,410 --> 00:12:53,010 Well, often what's happening here 352 352 00:12:53,010 --> 00:12:56,560 is there isn't necessarily going to be 100% utilization 353 353 00:12:56,560 --> 00:12:58,940 from router four to router five. 354 354 00:12:58,940 --> 00:13:01,110 Maybe that's our exterior one connection 355 355 00:13:01,110 --> 00:13:02,580 going out to the Internet. 356 356 00:13:02,580 --> 00:13:04,920 In fact, router one may only be sending you 357 357 00:13:04,920 --> 00:13:07,520 10 or 15 megabits per second right now. 358 358 00:13:07,520 --> 00:13:09,600 And router three might be sending you 30. 359 359 00:13:09,600 --> 00:13:11,700 And router six might be sending you one 360 360 00:13:11,700 --> 00:13:14,230 which is all added together less than 50. 361 361 00:13:14,230 --> 00:13:16,030 So, no buffering would occur. 362 362 00:13:16,030 --> 00:13:18,310 Now, there's also the possibility that router one 363 363 00:13:18,310 --> 00:13:20,910 and router three sends us more information. 364 364 00:13:20,910 --> 00:13:23,000 And that can cause buffering to occur 365 365 00:13:23,000 --> 00:13:25,610 because router four can't send enough data out 366 366 00:13:25,610 --> 00:13:28,770 to router five over that 50 megabit per second connection. 367 367 00:13:28,770 --> 00:13:31,270 So, this is the idea, we want to buffer things and hold it 368 368 00:13:31,270 --> 00:13:34,280 and then as we have room, we can send that information out, 369 369 00:13:34,280 --> 00:13:36,148 clear the buffer, and keep on moving. 370 370 00:13:36,148 --> 00:13:38,610 Because when we look at your networks, 371 371 00:13:38,610 --> 00:13:41,470 the chances are not every device is communicating 372 372 00:13:41,470 --> 00:13:44,300 to 100% of its capability all of the time. 373 373 00:13:44,300 --> 00:13:47,370 And so, by doing this, we can pay for a smaller connection 374 374 00:13:47,370 --> 00:13:50,300 to the outside world that 50 megabit connection, 375 375 00:13:50,300 --> 00:13:51,870 as opposed to paying for a large 376 376 00:13:51,870 --> 00:13:55,230 and expensive fiber connection of one gigabit per second. 377 377 00:13:55,230 --> 00:13:56,720 So, we can keep our costs down 378 378 00:13:56,720 --> 00:13:59,370 by knowing what our utilization is over our network 379 379 00:13:59,370 --> 00:14:00,850 for the long period of time. 380 380 00:14:00,850 --> 00:14:02,810 Now, that gets into some more advanced concepts 381 381 00:14:02,810 --> 00:14:04,600 that as you work as a network engineer, 382 382 00:14:04,600 --> 00:14:06,330 you'll start working on those designs 383 383 00:14:06,330 --> 00:14:08,170 and as bandwidth keeps getting cheaper, 384 384 00:14:08,170 --> 00:14:10,290 it becomes less and less important for us, 385 385 00:14:10,290 --> 00:14:12,840 at least in the small office, home office environment. 386 386 00:14:12,840 --> 00:14:15,363 But in large corporations, this is a big deal. 387 387 00:14:15,363 --> 00:14:19,180 Now, what are some examples of layer 4 devices? 388 388 00:14:19,180 --> 00:14:23,060 Well, we have TCP and UDP as our protocols for layer 4. 389 389 00:14:23,060 --> 00:14:24,840 So, if you see TCP and UDP, 390 390 00:14:24,840 --> 00:14:26,423 you know you're dealing with layer 4. 391 391 00:14:26,423 --> 00:14:29,000 We also have things like WAN accelerators, 392 392 00:14:29,000 --> 00:14:31,550 where we try to add compression to our IP packets, 393 393 00:14:31,550 --> 00:14:33,480 and then we send those segments over 394 394 00:14:33,480 --> 00:14:35,020 through those WAN accelerators 395 395 00:14:35,020 --> 00:14:37,110 to get them through our network faster. 396 396 00:14:37,110 --> 00:14:39,280 We also have load balancers and firewalls 397 397 00:14:39,280 --> 00:14:40,810 that can operate at layer 4 398 398 00:14:40,810 --> 00:14:43,440 by blocking and allowing different ports and protocols 399 399 00:14:43,440 --> 00:14:44,360 to go through them. 400 400 00:14:44,360 --> 00:14:46,520 For example, if you've ever gone in your firewall 401 401 00:14:46,520 --> 00:14:49,800 and blocked a port like Port 80 over TCP, 402 402 00:14:49,800 --> 00:14:51,770 well, that is a layer 4 block 403 403 00:14:51,770 --> 00:14:53,760 because you're blocking the port, Port 80, 404 404 00:14:53,760 --> 00:14:56,830 which is web traffic and the protocol TCP, 405 405 00:14:56,830 --> 00:14:58,853 which is our protocol at layer 4.