WEBVTT 1 00:00.000 --> 00:03.240 So, let's get back to the campus. 2 00:03.240 --> 00:07.710 We know that in a campus environment we're going to have a distribution layer 3 00:07.710 --> 00:07.880 and the 4 00:07.880 --> 00:11.900 distribution layer or distribution block is going to have a pair of switches 5 00:11.900 --> 00:12.120 and then 6 00:12.120 --> 00:17.960 we're going to bring those connections into the access layer in some capacity. 7 00:17.960 --> 00:22.760 So the access layer as we talked about earlier might be a single chassis, a 8 00:22.760 --> 00:23.720 single stack 9 00:23.720 --> 00:24.720 of switches. 10 00:24.720 --> 00:27.800 It could be a loop of switches that's hanging off of here. 11 00:27.800 --> 00:32.360 We could even end up potentially looping our network through and up to the WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 11 00:27.800 --> 00:32.360 We could even end up potentially looping our network through and up to the 12 00:32.360 --> 00:33.340 distribution 13 00:33.340 --> 00:34.340 layer. 14 00:34.340 --> 00:37.540 But either way, we know that we're going to come down to the access closet and 15 00:37.540 --> 00:37.880 one of 16 00:37.880 --> 00:43.360 these connections is going to get blocked by a spanning tree protocol. 17 00:43.360 --> 00:48.410 Now the distribution switch is going to have to maintain spanning tree root 18 00:48.410 --> 00:49.760 status up here 19 00:49.760 --> 00:54.420 in order to make sure that the access layer is actually blocking our links 20 00:54.420 --> 00:55.280 because the 21 00:55.280 --> 00:58.890 last thing we'd want is for the root to show up down here in one of our many 22 00:58.890 --> 00:59.600 access layer 23 00:59.600 --> 01:00.600 closets. WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 23 00:59.600 --> 01:00.600 closets. 24 01:00.600 --> 01:04.500 And now the distribution layer has to block links and it gets really awkward at 25 01:04.500 --> 01:05.120 that point 26 01:05.120 --> 01:08.560 when we're trying to troubleshoot network issues if the root isn't where we 27 01:08.560 --> 01:08.960 think it 28 01:08.960 --> 01:09.960 is. 29 01:09.960 --> 01:12.690 So we need to make sure that the spanning tree root exists up here at the 30 01:12.690 --> 01:13.440 distribution 31 01:13.440 --> 01:14.540 layer. 32 01:14.540 --> 01:18.880 We also know that we're going to deploy first-top redundancy protocols. 33 01:18.880 --> 01:22.790 And so the first-top redundancy protocols are going to result in us having 34 01:22.790 --> 01:23.560 active and 35 01:23.560 --> 01:25.360 standby connections. 36 01:25.360 --> 01:29.600 So these are some considerations that we have to make with our distribution 37 01:29.600 --> 01:31.000 layer configuration. WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 37 01:29.600 --> 01:31.000 layer configuration. 38 01:31.000 --> 01:35.480 But the question is how do we go about deploying these configurations to create 39 01:35.480 --> 01:36.240 an efficient 40 01:36.240 --> 01:37.400 network? 41 01:37.400 --> 01:43.620 For example, the classic scenario is that I make this left switch the root for 42 01:43.620 --> 01:44.760 all VLANs. 43 01:44.760 --> 01:49.880 If it's the root for all the VLANs, then we're going to block this link and use 44 01:49.880 --> 01:50.600 this link 45 01:50.600 --> 01:53.200 for all of our connections. 46 01:53.200 --> 01:55.360 And so our topology is going to look like this. 47 01:55.360 --> 02:01.200 Meanwhile, this gig or 10-gig connection here is completely idle. WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 47 01:55.360 --> 02:01.200 Meanwhile, this gig or 10-gig connection here is completely idle. 48 02:01.200 --> 02:03.920 And we want to prevent this if at all possible. 49 02:03.920 --> 02:07.820 We've spent a lot of money for that 10-gig connection and those 10-gig optics 50 02:07.820 --> 02:09.080 and the fiber 51 02:09.080 --> 02:10.080 that's required. 52 02:10.080 --> 02:15.330 And so how can we go about making it so that we at least use that 10-gig link 53 02:15.330 --> 02:16.040 as often 54 02:16.040 --> 02:17.640 as we can? 55 02:17.640 --> 02:23.440 Now this is going to lead us to the conversation of Cisco's per VLAN spanning 56 02:23.440 --> 02:25.120 tree and ultimately 57 02:25.120 --> 02:28.440 our ability to deploy policies per VLAN. 58 02:28.440 --> 02:35.390 This is going to apply on a basis of spanning tree, whether we're running Cisco WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 58 02:28.440 --> 02:35.390 This is going to apply on a basis of spanning tree, whether we're running Cisco 59 02:35.390 --> 02:37.040 's PVST+ or 60 02:37.040 --> 02:44.840 rapid PVST+ and even the industry standard multiple spanning tree. 61 02:44.840 --> 02:50.680 We can deploy per VLAN topologies with STP and therefore I could create a root 62 02:50.680 --> 02:51.520 switch over 63 02:51.520 --> 02:53.240 here for another VLAN. 64 02:53.240 --> 02:58.760 So let's say the left root is for VLAN 10 and the right root is for VLAN 20. 65 02:58.760 --> 03:03.980 Well now VLAN 20 will block this link on the left and will allow traffic to WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 65 02:58.760 --> 03:03.980 Well now VLAN 20 will block this link on the left and will allow traffic to 66 03:03.980 --> 03:04.720 flow like 67 03:04.720 --> 03:06.200 this. 68 03:06.200 --> 03:08.840 So now we have two different topologies. 69 03:08.840 --> 03:10.600 One looks like this. 70 03:10.600 --> 03:14.120 The other looks like, well, this. 71 03:14.120 --> 03:16.040 There we go. 72 03:16.040 --> 03:18.920 And as a result, we are using both of those links. 73 03:18.920 --> 03:24.320 We're just making it so VLAN 10 can't use the link on the right and VLAN 20 can 74 03:24.320 --> 03:24.880 't use 75 03:24.880 --> 03:30.320 the link on the left that prevents loops and it prevents the concept of one of WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 75 03:24.880 --> 03:30.320 the link on the left that prevents loops and it prevents the concept of one of 76 03:30.320 --> 03:30.920 the links 77 03:30.920 --> 03:34.040 being idle, which is our primary goal. 78 03:34.040 --> 03:39.040 But the per VLAN concept also applies to our first top redundancy protocols. 79 03:39.040 --> 03:46.140 So whether we're talking about HSRP or VRRP, especially since GLBP will low 80 03:46.140 --> 03:46.640 balance, we're 81 03:46.640 --> 03:52.360 going to be setting our active and our standby according to the VLAN. 82 03:52.360 --> 03:58.170 I can make it so that on VLAN 10, let's just go back to our example on the left 83 03:58.170 --> 03:58.720 that for 84 03:58.720 --> 04:02.920 VLAN 10 we're also the HSRP active. WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 84 03:58.720 --> 04:02.920 VLAN 10 we're also the HSRP active. 85 04:02.920 --> 04:08.520 But for VLAN 20, the device on the right is HSRP active. 86 04:08.520 --> 04:13.360 This way, we can stay very efficient with our design. 87 04:13.360 --> 04:18.440 Think back to what we just said about VPCs with efficient traffic forwarding. 88 04:18.440 --> 04:26.140 Imagine for a moment that it's not the right switch that's active for VLAN 20, 89 04:26.140 --> 04:26.560 but the left 90 04:26.560 --> 04:29.800 switch is also active for VLAN 20. 91 04:29.800 --> 04:33.400 So what happens with the VLAN 20 when a packet arrives here that's destined for WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 91 04:29.800 --> 04:33.400 So what happens with the VLAN 20 when a packet arrives here that's destined for 92 04:33.400 --> 04:34.440 the gateway? 93 04:34.440 --> 04:39.240 Well, it's going to be forwarded up here to the root because this link is 94 04:39.240 --> 04:39.960 blocked. 95 04:39.960 --> 04:43.900 And then it's going to have to get forwarded across the link to gain access to 96 04:43.900 --> 04:44.520 the SVI over 97 04:44.520 --> 04:48.920 there because that's where the HSRP active is. 98 04:48.920 --> 04:54.040 We want to avoid having that inefficient communication path. 99 04:54.040 --> 04:59.980 And so instead, we are going to apply the active configuration to wherever the 100 04:59.980 --> 05:00.640 root is WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 100 04:59.980 --> 05:00.640 root is 101 05:00.640 --> 05:02.960 on a per VLAN basis. 102 05:02.960 --> 05:06.280 So that is the primary takeaway from this conversation. 103 05:06.280 --> 05:13.050 When we're looking at aligning the configurations, maybe of the spanning tree 104 05:13.050 --> 05:15.520 root and the FHRP's, 105 05:15.520 --> 05:26.940 we want to make sure that on a per VLAN basis that the active and the root are 106 05:26.940 --> 05:28.360 the same. 107 05:28.360 --> 05:32.840 And again, that could be the BRRP backup as well. WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 107 05:28.360 --> 05:32.840 And again, that could be the BRRP backup as well. 108 05:32.840 --> 05:36.240 So this is our FHRP configuration. 109 05:36.240 --> 05:39.000 This is our STP configuration. 110 05:39.000 --> 05:42.520 And usually we want to split this up evenly across our VLANs. 111 05:42.520 --> 05:50.880 So maybe I have VLANs 10, 20, 30, and 40 in my environment. 112 05:50.880 --> 05:55.520 Well, I'm going to take VLANs 10 and 30. 113 05:55.520 --> 06:02.020 And I'm going to make, let's say, this switch on the left, both the FHRP active WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 113 05:55.520 --> 06:02.020 And I'm going to make, let's say, this switch on the left, both the FHRP active 114 06:02.020 --> 06:02.400 and the 115 06:02.400 --> 06:03.400 STP root. 116 06:03.400 --> 06:09.360 But for VLANs 20 and 40, I'm going to go to the switch on the right. 117 06:09.360 --> 06:12.920 And I've just divided my network up by VLAN. 118 06:12.920 --> 06:18.270 Now a lot of networks choose to split us this up by evens and odds or something 119 06:18.270 --> 06:19.040 along those 120 06:19.040 --> 06:20.040 lines. 121 06:20.040 --> 06:21.720 Now all four of those are even numbers. 122 06:21.720 --> 06:27.860 But if I look at the tens digit here, one in three, these would be my odd VLANs 123 06:27.860 --> 06:28.200 . 124 06:28.200 --> 06:32.040 And these would be my even VLANs over on the right. WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 124 06:28.200 --> 06:32.040 And these would be my even VLANs over on the right. 125 06:32.040 --> 06:36.930 If I actually have VLANs 1 through 9, something like that, then I could 126 06:36.930 --> 06:38.480 actually choose true, 127 06:38.480 --> 06:40.280 odd, and even values. 128 06:40.280 --> 06:45.060 But ultimately, in a lot of situations, we're looking more to keep the number 129 06:45.060 --> 06:45.880 the same. 130 06:45.880 --> 06:51.480 We want to balance by VLAN count. 131 06:51.480 --> 06:54.530 And this is just from an organizational perspective, but it does make an 132 06:54.530 --> 06:55.040 underlying 133 06:55.040 --> 06:56.280 assumption. 134 06:56.280 --> 07:04.120 And that is that different VLANs have roughly the same traffic flow. WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 134 06:56.280 --> 07:04.120 And that is that different VLANs have roughly the same traffic flow. 135 07:04.120 --> 07:08.140 This is very common in most scenarios where we have endpoints that are just 136 07:08.140 --> 07:08.880 going out to 137 07:08.880 --> 07:12.720 the internet or to data center servers. 138 07:12.720 --> 07:16.560 Because HTTP traffic just doesn't take up that much, we can look at our 10-gag 139 07:16.560 --> 07:17.120 links and 140 07:17.120 --> 07:21.280 realize that most of them are highly underutilized in a campus environment. 141 07:21.280 --> 07:27.360 But if I do happen to have a heavy load on VLANs 10 and 30, these are very, 142 07:27.360 --> 07:28.280 very heavy loads 143 07:28.280 --> 07:32.970 for some reason, and I place them onto the same distribution switch, well, at WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 143 07:28.280 --> 07:32.970 for some reason, and I place them onto the same distribution switch, well, at 144 07:32.970 --> 07:33.840 that point, 145 07:33.840 --> 07:36.520 I'm heavily loading up one of my links. 146 07:36.520 --> 07:39.600 And so that would be something to consider as well. 147 07:39.600 --> 07:43.190 But usually in campus environments in the real world, we don't need to think 148 07:43.190 --> 07:44.040 too hard about 149 07:44.040 --> 07:45.040 that. 150 07:45.040 --> 07:48.730 Now, one last caveat to consider, and yes, this is me bringing up the data 151 07:48.730 --> 07:50.120 center again, 152 07:50.120 --> 07:55.020 but we just looked at how in the data center we're going to run VPC 153 07:55.020 --> 07:56.560 environments. 154 07:56.560 --> 08:01.890 When we have VPC peers, we just got done saying that in a VPC environment, let WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 154 07:56.560 --> 08:01.890 When we have VPC peers, we just got done saying that in a VPC environment, let 155 08:01.890 --> 08:02.760 's draw that 156 08:02.760 --> 08:07.750 out real quick, there we go, that we are going to load balance traffic both 157 08:07.750 --> 08:08.040 ways. 158 08:08.040 --> 08:12.960 And we have active active up here on our FHRP. 159 08:12.960 --> 08:18.160 This completely changes the game. We have active active at the FHRP level, and 160 08:18.160 --> 08:22.000 we have no blocking 161 08:22.000 --> 08:24.280 from an STP perspective. 162 08:24.280 --> 08:29.110 So at this point, we have to ask the question, does it matter which switch is 163 08:29.110 --> 08:30.520 which from a root WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 163 08:29.110 --> 08:30.520 which from a root 164 08:30.520 --> 08:33.520 and FHRP active alignment? 165 08:33.520 --> 08:36.520 And the answer is it doesn't really matter anymore. 166 08:36.520 --> 08:39.760 As a result, Cisco changes the recommendations. 167 08:39.760 --> 08:46.120 Our VPC peers are going to have a primary and secondary role. 168 08:46.120 --> 08:51.170 That goes beyond the scope of the course, but in this case, we can identify 169 08:51.170 --> 08:52.280 that this is 170 08:52.280 --> 08:57.960 the VPC primary, and Cisco recommends that if it's the VPC primary, that we go 171 08:57.960 --> 08:58.440 ahead 172 08:58.440 --> 09:09.360 and make it the STP root, and the FHRP active for all VLANs. WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 172 08:58.440 --> 09:09.360 and make it the STP root, and the FHRP active for all VLANs. 173 09:09.360 --> 09:13.320 We're not fascinating how that recommendation changes. 174 09:13.320 --> 09:19.440 And the purpose here is just mostly for us to be able to recognize that it's an 175 09:19.440 --> 09:20.480 arbitrary 176 09:20.480 --> 09:23.320 configuration, it doesn't really matter how we do it. 177 09:23.320 --> 09:26.970 If we choose to split it up still, we're not causing any harm, but the reason 178 09:26.970 --> 09:27.600 why Cisco 179 09:27.600 --> 09:31.400 does this is so that we can look at our environment and we can know exactly how WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 179 09:27.600 --> 09:31.400 does this is so that we can look at our environment and we can know exactly how 180 09:31.400 --> 09:32.720 it's running 181 09:32.720 --> 09:35.200 from a control plane perspective. 182 09:35.200 --> 09:38.940 And because the control plane is very lightly loading, we don't need to worry 183 09:38.940 --> 09:40.040 about overloading 184 09:40.040 --> 09:44.800 one of the switches, and this way we know exactly how the network will behave 185 09:44.800 --> 09:45.520 if we lose 186 09:45.520 --> 09:47.400 one of these devices. 187 09:47.400 --> 09:52.120 So that's why the recommendations from Cisco change in the data center when we 188 09:52.120 --> 09:52.640 're running 189 09:52.640 --> 09:55.800 VPC peers at the aggregation layer. 190 09:55.800 --> 10:01.220 But outside of that, here at our campus environment, with a true distribution WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 190 09:55.800 --> 10:01.220 But outside of that, here at our campus environment, with a true distribution 191 10:01.220 --> 10:01.960 layer, we're 192 10:01.960 --> 10:06.970 going to want to make sure that we are splitting up the STP, well, I should say 193 10:06.970 --> 10:08.000 like this, we 194 10:08.000 --> 10:15.490 are aligning our STP roots with our FHRP active right here, but we're splitting 195 10:15.490 --> 10:16.320 that up on 196 10:16.320 --> 10:18.320 a perv LAN basis. 197 10:18.320 --> 10:21.400 That will be the goal for us when we're performing our campus designs.