1 1 00:00:00,600 --> 00:00:02,270 The last topic we need to talk about 2 2 00:00:02,270 --> 00:00:05,790 in this routing section is how multicast routing works. 3 3 00:00:05,790 --> 00:00:08,080 Now, multicast routing is when you send traffic 4 4 00:00:08,080 --> 00:00:12,020 to a class D IP address, known as the multicast group. 5 5 00:00:12,020 --> 00:00:15,050 Now, our goal here is for us only to send traffic out once 6 6 00:00:15,050 --> 00:00:16,240 and then have all of the devices 7 7 00:00:16,240 --> 00:00:18,320 who want to get that information, 8 8 00:00:18,320 --> 00:00:21,120 and anybody who doesn't want to get it, doesn't get it. 9 9 00:00:21,120 --> 00:00:23,520 Now, there are two primary ways of doing this. 10 10 00:00:23,520 --> 00:00:25,510 The first is known as IGMP, 11 11 00:00:25,510 --> 00:00:27,840 or the Internet Group Management Protocol. 12 12 00:00:27,840 --> 00:00:31,980 The second is PIM, or Protocol Independent Multicast. 13 13 00:00:31,980 --> 00:00:35,150 We'll cover both of those in detail in this lesson. 14 14 00:00:35,150 --> 00:00:38,310 Now, IGMP, or the Internet Group Management Protocol, 15 15 00:00:38,310 --> 00:00:39,930 is used by clients and routers 16 16 00:00:39,930 --> 00:00:41,730 so that the routers know which interfaces 17 17 00:00:41,730 --> 00:00:43,603 have multicast receivers on them. 18 18 00:00:43,603 --> 00:00:45,955 This allows a client to join a multicast group 19 19 00:00:45,955 --> 00:00:48,750 and then be able to receive messages that they wanted to get 20 20 00:00:48,750 --> 00:00:50,380 through this multicast group. 21 21 00:00:50,380 --> 00:00:52,900 Now, there are three variants of IGMP. 22 22 00:00:52,900 --> 00:00:55,540 The first one was IGMPv1. 23 23 00:00:55,540 --> 00:00:58,530 And then it was improved to version two and version three. 24 24 00:00:58,530 --> 00:00:59,650 Now in version one, 25 25 00:00:59,650 --> 00:01:01,310 clients could request joining the group 26 26 00:01:01,310 --> 00:01:04,510 and every 60 seconds, their router would go back and say, 27 27 00:01:04,510 --> 00:01:06,493 do you still want to be here? Do you still want to be here? 28 28 00:01:06,493 --> 00:01:07,697 Do you still want to be here? 29 29 00:01:07,697 --> 00:01:09,020 And you could see how that causes 30 30 00:01:09,020 --> 00:01:10,910 a lot of unnecessary traffic. 31 31 00:01:10,910 --> 00:01:12,910 In version two, clients had the ability 32 32 00:01:12,910 --> 00:01:16,070 to send leave messages to exit the group when they wanted. 33 33 00:01:16,070 --> 00:01:18,120 Essentially now, the router's just going to assume 34 34 00:01:18,120 --> 00:01:19,700 you want to be there until you told it 35 35 00:01:19,700 --> 00:01:21,290 that you didn't want to be there. 36 36 00:01:21,290 --> 00:01:22,870 Finally, in version three, 37 37 00:01:22,870 --> 00:01:24,560 the client could request a multicast 38 38 00:01:24,560 --> 00:01:27,040 from only a specific server and then choose, 39 39 00:01:27,040 --> 00:01:29,060 did I want to get my messages from one server 40 40 00:01:29,060 --> 00:01:30,500 and not from server two? 41 41 00:01:30,500 --> 00:01:31,610 That was allowed. 42 42 00:01:31,610 --> 00:01:34,130 It would also allow for source specific multicast, 43 43 00:01:34,130 --> 00:01:36,340 which is what this server request was called. 44 44 00:01:36,340 --> 00:01:38,610 Now, this allowed us to have multiple video streams 45 45 00:01:38,610 --> 00:01:40,810 to a single multicast stream. 46 46 00:01:40,810 --> 00:01:42,810 Now, almost like you'd have with cable TV, 47 47 00:01:42,810 --> 00:01:44,790 where you could dial into the channel you wanted, 48 48 00:01:44,790 --> 00:01:46,760 that's kind of what we were doing here digitally, 49 49 00:01:46,760 --> 00:01:48,980 using the IGMP protocol. 50 50 00:01:48,980 --> 00:01:51,220 So how did IGMP work? 51 51 00:01:51,220 --> 00:01:53,010 Well, let's look at this diagram. 52 52 00:01:53,010 --> 00:01:55,420 If I have a server that wants to send out traffic, 53 53 00:01:55,420 --> 00:01:56,930 it's going to send it to the router 54 54 00:01:56,930 --> 00:01:59,480 and the IP address of the multicast group. 55 55 00:01:59,480 --> 00:02:03,600 In this case, 239.2.1.3. 56 56 00:02:03,600 --> 00:02:05,970 Right now, the router is not sending it anywhere, 57 57 00:02:05,970 --> 00:02:08,290 'cause nobody's requested that information. 58 58 00:02:08,290 --> 00:02:11,720 But if PC2 sends a request message to router one, 59 59 00:02:11,720 --> 00:02:13,945 now router one is going to remember that PC2 60 60 00:02:13,945 --> 00:02:16,094 wants to be a part of this multicast group, 61 61 00:02:16,094 --> 00:02:18,720 and anything it gets as part of this multicast group 62 62 00:02:18,720 --> 00:02:21,300 will now be sent over to PC2 as well. 63 63 00:02:21,300 --> 00:02:22,850 And only going to go to PC2, 64 64 00:02:22,850 --> 00:02:25,270 because remember, PC1 and PC3 65 65 00:02:25,270 --> 00:02:26,932 didn't ask to be a part of this group. 66 66 00:02:26,932 --> 00:02:29,430 So hopefully you can see the benefits of this now. 67 67 00:02:29,430 --> 00:02:30,550 In this example here, 68 68 00:02:30,550 --> 00:02:34,050 we have PC1 and PC3 that wanted to be in there 69 69 00:02:34,050 --> 00:02:35,730 and they could join the group as well. 70 70 00:02:35,730 --> 00:02:38,500 And then the server can send it out to all three people 71 71 00:02:38,500 --> 00:02:41,200 with only one copy of the message being sent in. 72 72 00:02:41,200 --> 00:02:42,636 That's the benefit of this. 73 73 00:02:42,636 --> 00:02:44,880 The second thing we want to look at is PIM, 74 74 00:02:44,880 --> 00:02:47,100 or the Protocol Independent Multicast. 75 75 00:02:47,100 --> 00:02:49,890 PIM is going to allow multicast traffic to be routed 76 76 00:02:49,890 --> 00:02:52,110 between multicast enabled routers. 77 77 00:02:52,110 --> 00:02:55,360 Multicast routing forms this multicast distribution tree, 78 78 00:02:55,360 --> 00:02:57,550 and it really works between all these different routers 79 79 00:02:57,550 --> 00:02:59,360 together because IGMP 80 80 00:02:59,360 --> 00:03:01,320 was more about clients and servers together, 81 81 00:03:01,320 --> 00:03:04,880 where PIM is much more focused on the routing part of this. 82 82 00:03:04,880 --> 00:03:07,180 Now, there are two different modes in PIM. 83 83 00:03:07,180 --> 00:03:09,500 There's a PIM-DM, which is the Dense Mode, 84 84 00:03:09,500 --> 00:03:12,010 or PIM-SM or Sparse Mode. 85 85 00:03:12,010 --> 00:03:14,950 In Dense Mode, we're going to use a periodic flood and prune 86 86 00:03:14,950 --> 00:03:17,520 behavior to form an optimal distribution tree 87 87 00:03:17,520 --> 00:03:19,070 across these routers. 88 88 00:03:19,070 --> 00:03:21,400 And this can actually cause a negative performance impact 89 89 00:03:21,400 --> 00:03:24,000 on your network, and because of this, in modern networks, 90 90 00:03:24,000 --> 00:03:25,860 we really don't use it. 91 91 00:03:25,860 --> 00:03:28,160 Instead, we tend to use Sparse Mode. 92 92 00:03:28,160 --> 00:03:30,280 With PIM-SM, or Sparse Mode, 93 93 00:03:30,280 --> 00:03:32,960 this is going to use a shared distribution tree initially, 94 94 00:03:32,960 --> 00:03:36,380 and then it's going to, over time, find the best tree. 95 95 00:03:36,380 --> 00:03:37,490 Now, when we start out, 96 96 00:03:37,490 --> 00:03:40,980 that shared distribution tree is not optimal, but over time, 97 97 00:03:40,980 --> 00:03:42,720 we're going to learn where the best tree is, 98 98 00:03:42,720 --> 00:03:45,350 and then start switching over to the Shortest Path Tree, 99 99 00:03:45,350 --> 00:03:49,340 or SPT, once it determines what that actually is. 100 100 00:03:49,340 --> 00:03:50,900 Now, I know that's a lot of words. 101 101 00:03:50,900 --> 00:03:52,530 So let's take a look at some pictures 102 102 00:03:52,530 --> 00:03:55,020 to hopefully understand how this works a little better. 103 103 00:03:55,020 --> 00:03:58,090 Here you can see PIM-DM or Dense Mode. 104 104 00:03:58,090 --> 00:03:59,780 Now, what happens in Dense Mode is, 105 105 00:03:59,780 --> 00:04:02,260 to begin with, it has this flooding procedure 106 106 00:04:02,260 --> 00:04:03,640 where all of the routers, 107 107 00:04:03,640 --> 00:04:05,150 every router in this entire network, 108 108 00:04:05,150 --> 00:04:07,030 is getting all of the information. 109 109 00:04:07,030 --> 00:04:08,990 Now, as you can see, that's a lot of traffic, 110 110 00:04:08,990 --> 00:04:11,780 but we do already have the optimal way of getting there 111 111 00:04:11,780 --> 00:04:15,270 from that multicast source to the multicast destination, 112 112 00:04:15,270 --> 00:04:18,690 because it comes all at once with that big flood. 113 113 00:04:18,690 --> 00:04:21,040 So in this case, it would be going down the left side 114 114 00:04:21,040 --> 00:04:22,310 of that pyramid. 115 115 00:04:22,310 --> 00:04:23,380 Now, as we go through, 116 116 00:04:23,380 --> 00:04:26,010 it's going to then prune off all the non-optimal routes, 117 117 00:04:26,010 --> 00:04:28,690 which in this case is all the routers off to the right. 118 118 00:04:28,690 --> 00:04:30,270 It sends out this prune message saying, 119 119 00:04:30,270 --> 00:04:31,980 hey, I don't need this traffic. 120 120 00:04:31,980 --> 00:04:34,800 I'm not part of the optical route. Get me out of here. 121 121 00:04:34,800 --> 00:04:35,633 And that way, 122 122 00:04:35,633 --> 00:04:37,800 they can go ahead and drop those from the route. 123 123 00:04:37,800 --> 00:04:39,870 Now, after sending all these pruning messages, 124 124 00:04:39,870 --> 00:04:41,230 we now have the optimal path 125 125 00:04:41,230 --> 00:04:44,000 between the source router and the last hop router, 126 126 00:04:44,000 --> 00:04:46,690 giving us the quickest and easiest path. 127 127 00:04:46,690 --> 00:04:48,920 Now, this is the idea of flood and prune. 128 128 00:04:48,920 --> 00:04:50,800 And this happens every three minutes, 129 129 00:04:50,800 --> 00:04:52,570 trying to find a more optimal route, 130 130 00:04:52,570 --> 00:04:54,980 which again, can cause a really big performance hit 131 131 00:04:54,980 --> 00:04:56,790 to your network, because we're flooding it 132 132 00:04:56,790 --> 00:04:58,200 every three minutes. 133 133 00:04:58,200 --> 00:05:00,640 And this is why we don't really use it much anymore. 134 134 00:05:00,640 --> 00:05:02,940 Instead we use Sparse Mode. 135 135 00:05:02,940 --> 00:05:05,340 Now Sparse Mode, or PIM-SM, 136 136 00:05:05,340 --> 00:05:07,710 is going to use a shared distribution tree. 137 137 00:05:07,710 --> 00:05:10,360 Essentially, when the server sends out the first message, 138 138 00:05:10,360 --> 00:05:12,610 the router's going to send it over any way it can 139 139 00:05:12,610 --> 00:05:15,700 to get it from the first router down to the last hop. 140 140 00:05:15,700 --> 00:05:17,260 In this case, it's going to be sending it 141 141 00:05:17,260 --> 00:05:19,010 to this rendezvous point first, 142 142 00:05:19,010 --> 00:05:21,230 and then down to last router hop. 143 143 00:05:21,230 --> 00:05:22,910 Now, this isn't the optimal path, 144 144 00:05:22,910 --> 00:05:25,340 but it is a path and that's okay and it works. 145 145 00:05:25,340 --> 00:05:27,840 And so this information is going to start to flow. 146 146 00:05:27,840 --> 00:05:28,800 As you can see, 147 147 00:05:28,800 --> 00:05:31,780 there are four of the six routers that got the information. 148 148 00:05:31,780 --> 00:05:34,130 Two of them weren't even bothered to get the information, 149 149 00:05:34,130 --> 00:05:35,900 and so they saved all that resources, 150 150 00:05:35,900 --> 00:05:37,580 because they weren't being flooded. 151 151 00:05:37,580 --> 00:05:40,630 Now, over time, we figured out this was a sub optimal path, 152 152 00:05:40,630 --> 00:05:42,260 and instead, it would be a lot quicker 153 153 00:05:42,260 --> 00:05:44,380 to go down the left-hand side of the diagram. 154 154 00:05:44,380 --> 00:05:46,370 So, we're going to start switching over to that, 155 155 00:05:46,370 --> 00:05:48,250 getting rid of all these unused branches 156 156 00:05:48,250 --> 00:05:51,340 and then going to the Shortest Path Tree, the SPT, 157 157 00:05:51,340 --> 00:05:53,160 and that now gives us the optimal tree 158 158 00:05:53,160 --> 00:05:54,860 in this shared distribution tree, 159 159 00:05:54,860 --> 00:05:56,890 just like we had before in Dense Mode. 160 160 00:05:56,890 --> 00:05:59,160 Now you can see how this uses a lot less resources 161 161 00:05:59,160 --> 00:06:00,390 than Dense Mode did, 162 162 00:06:00,390 --> 00:06:02,420 but it did take us a little bit more time 163 163 00:06:02,420 --> 00:06:04,210 to find that optimal way. 164 164 00:06:04,210 --> 00:06:05,700 That's the trade off here. 165 165 00:06:05,700 --> 00:06:07,700 Do you want to get optimal right from the start, 166 166 00:06:07,700 --> 00:06:09,090 but flood your network? 167 167 00:06:09,090 --> 00:06:10,490 Or can you wait for optimal 168 168 00:06:10,490 --> 00:06:12,220 and use a lot less resources? 169 169 00:06:12,220 --> 00:06:13,630 In most modern networks, 170 170 00:06:13,630 --> 00:06:16,170 we choose PIM-SM, or Sparse Mode, 171 171 00:06:16,170 --> 00:06:17,890 because it works better on our networks 172 172 00:06:17,890 --> 00:06:19,650 and uses less resources upfront, 173 173 00:06:19,650 --> 00:06:21,760 and eventually gets that optimal distribution 174 174 00:06:21,760 --> 00:06:22,910 that we're looking for.