WEBVTT 1 00:00.000 --> 00:06.320 So we can solve all of our layer two problems by deploying layer three to the 2 00:06.320 --> 00:08.400 access layer. 3 00:08.400 --> 00:13.160 This is a pretty interesting way to deploy our access layer devices. 4 00:13.160 --> 00:16.920 So same thing, we start with a pair of distribution switches. 5 00:16.920 --> 00:20.170 But now we can connect them at layer three to one another because what we're 6 00:20.170 --> 00:21.240 going to do 7 00:21.240 --> 00:24.640 is come down to an access closet here. 8 00:24.640 --> 00:28.040 And let's say we have a stack of switches or chassis switch. 9 00:28.040 --> 00:34.920 I'm going to extend layer three down to that access closet. WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 9 00:28.040 --> 00:34.920 I'm going to extend layer three down to that access closet. 10 00:34.920 --> 00:37.360 This means that I am now performing routing. 11 00:37.360 --> 00:39.960 I'm going to have to have a routing protocol running here. 12 00:39.960 --> 00:43.600 Maybe it's OSPF or EIGRP. 13 00:43.600 --> 00:48.010 And by creating this as a layer three link, this means that there is no longer 14 00:48.010 --> 00:48.720 any spanning 15 00:48.720 --> 00:51.600 tree protocol needed. 16 00:51.600 --> 00:57.470 Furthermore, because we still need VLANs and default gateways and everything, 17 00:57.470 --> 00:58.320 now our 18 00:58.320 --> 01:04.170 closet switch here is going to be responsible for defining the VLANs and the SV WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 18 00:58.320 --> 01:04.170 closet switch here is going to be responsible for defining the VLANs and the SV 19 01:04.170 --> 01:04.680 Is. 20 01:04.680 --> 01:10.140 So that as our users connect in, they still have access to their default gate 21 01:10.140 --> 01:10.800 ways. 22 01:10.800 --> 01:13.980 The default gateways are just going to be here in the access closet at which 23 01:13.980 --> 01:14.320 point 24 01:14.320 --> 01:19.560 we get routed upstream to that layer three distribution layer still. 25 01:19.560 --> 01:25.520 So more if I deploy another closet, I'm most of our networks have many closets. 26 01:25.520 --> 01:27.480 We're going to have another switch here. 27 01:27.480 --> 01:29.120 We can bring layer three down. 28 01:29.120 --> 01:33.090 And again, since everything is layer three, we don't have any blocked links in WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 28 01:29.120 --> 01:33.090 And again, since everything is layer three, we don't have any blocked links in 29 01:33.090 --> 01:34.120 this environment. 30 01:34.120 --> 01:37.160 We again, though, have to create local VLANs. 31 01:37.160 --> 01:41.160 So I might have VLANs 10 through 15 on the left. 32 01:41.160 --> 01:42.160 But then guess what? 33 01:42.160 --> 01:46.240 I can have VLANs 10 through 15 on the right as well because the VLAN IDs are 34 01:46.240 --> 01:47.600 locally significant. 35 01:47.600 --> 01:49.160 So that's kind of interesting. 36 01:49.160 --> 01:52.710 The subnet still have to be unique, but the VLAN IDs themselves can be dupl 37 01:52.710 --> 01:53.280 icated, which 38 01:53.280 --> 01:56.360 is one less thing that I have to manage. 39 01:56.360 --> 02:00.570 Now do note that from a hierarchical model perspective, this does change things WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 39 01:56.360 --> 02:00.570 Now do note that from a hierarchical model perspective, this does change things 40 02:00.570 --> 02:01.000 a little 41 02:01.000 --> 02:06.300 bit because now our SVIs, which normally reside on the distribution layer, 42 02:06.300 --> 02:07.320 truly reside 43 02:07.320 --> 02:09.120 at the access layer. 44 02:09.120 --> 02:12.770 So just keep that in mind, it doesn't break the model, but it does slightly 45 02:12.770 --> 02:13.240 tweak the 46 02:13.240 --> 02:15.400 way that the model is deployed. 47 02:15.400 --> 02:19.460 And within the context of a layer three access layer, that's perfectly 48 02:19.460 --> 02:20.440 acceptable. 49 02:20.440 --> 02:24.120 So let's talk about the pros and cons of this design. 50 02:24.120 --> 02:29.430 The pros, absolutely the fact that we have no STP from spanning tree protocol, 51 02:29.430 --> 02:29.640 that is 52 02:29.640 --> 02:31.200 a huge win. WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 52 02:29.640 --> 02:31.200 a huge win. 53 02:31.200 --> 02:35.710 Anytime we can eliminate spanning tree protocol or specifically the need for it 54 02:35.710 --> 02:36.200 , that 55 02:36.200 --> 02:37.480 point it's a win. 56 02:37.480 --> 02:41.270 We shouldn't just get rid of spanning tree protocol on layer two network 57 02:41.270 --> 02:41.960 because that 58 02:41.960 --> 02:42.960 wouldn't be a win. 59 02:42.960 --> 02:45.040 That would create lots of loops. 60 02:45.040 --> 02:46.840 Here three though, we're good to go. 61 02:46.840 --> 02:52.150 We also can safely rely on the fact that we can scale this up to any number of 62 02:52.150 --> 02:53.720 access closets. 63 02:53.720 --> 02:57.910 I don't have to worry about the size of my broadcast domains anymore because 64 02:57.910 --> 02:58.440 those are 65 02:58.440 --> 03:02.280 completely restricted within each one of these closets. WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 65 02:58.440 --> 03:02.280 completely restricted within each one of these closets. 66 03:02.280 --> 03:05.920 This is a broadcast domain here. 67 03:05.920 --> 03:08.120 This is a broadcast domain here. 68 03:08.120 --> 03:11.600 Every single closet is its own little layer to domain. 69 03:11.600 --> 03:15.260 And so we don't have to worry about broadcast being sent out unnecessarily to 70 03:15.260 --> 03:15.880 all the other 71 03:15.880 --> 03:18.400 access closets in the network. 72 03:18.400 --> 03:22.200 So should we all go with layer three access layer deployments? 73 03:22.200 --> 03:26.930 Well maybe, but we should also stop and consider that there are some downsides 74 03:26.930 --> 03:28.080 to this design. 75 03:28.080 --> 03:33.520 First of all, layer three switches are more expensive. WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 75 03:28.080 --> 03:33.520 First of all, layer three switches are more expensive. 76 03:33.520 --> 03:37.280 And unfortunately, this is something we have to consider when it comes to our 77 03:37.280 --> 03:37.720 network 78 03:37.720 --> 03:39.000 designs. 79 03:39.000 --> 03:41.680 Most of us don't have infinite budgets. 80 03:41.680 --> 03:45.360 And so if I have to spend, just making up a number here, but if I have to spend 81 03:45.360 --> 03:45.960 twice as 82 03:45.960 --> 03:51.360 much for every single switch I deploy into my access layer and my access layer, 83 03:51.360 --> 03:51.880 I have 84 03:51.880 --> 03:52.880 lots of switches. 85 03:52.880 --> 03:57.120 Oh, that's going to double the size of my access layer budget. 86 03:57.120 --> 04:01.550 Or if it increases only by 30 or 40%, it still increases the overall budget by WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 86 03:57.120 --> 04:01.550 Or if it increases only by 30 or 40%, it still increases the overall budget by 87 04:01.550 --> 04:02.960 that amount. 88 04:02.960 --> 04:07.620 And so we have to be concerned about the amount of money that is going to cost 89 04:07.620 --> 04:08.160 and whether 90 04:08.160 --> 04:10.680 we can cost justify it. 91 04:10.680 --> 04:15.480 Furthermore, we do lose that flexibility that we talked about. 92 04:15.480 --> 04:20.130 On day one, I might not have anything that requires layer two connectivity 93 04:20.130 --> 04:20.960 between my 94 04:20.960 --> 04:22.040 closets. 95 04:22.040 --> 04:24.840 So hey, if we're good today, that's really great. 96 04:24.840 --> 04:30.320 But what happens if in the future I do end up needing to have an application or WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 96 04:24.840 --> 04:30.320 But what happens if in the future I do end up needing to have an application or 97 04:30.320 --> 04:30.760 maybe 98 04:30.760 --> 04:35.480 a set of devices that require layer two connectivity with one another. 99 04:35.480 --> 04:39.040 So this is potentially a concern. 100 04:39.040 --> 04:44.610 And the danger here is that maybe I don't need it today, but I might need it 101 04:44.610 --> 04:45.720 tomorrow. 102 04:45.720 --> 04:52.120 I have been part of a network project, a deployment issue where on day one, a 103 04:52.120 --> 04:53.520 network was deployed 104 04:53.520 --> 04:56.720 at layer three and that was perfectly fine. 105 04:56.720 --> 05:02.380 But later on, when I came into the mix, we were hired to help resolve this WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 105 04:56.720 --> 05:02.380 But later on, when I came into the mix, we were hired to help resolve this 106 05:02.380 --> 05:03.560 problem where 107 05:03.560 --> 05:07.850 now they needed devices on both sides of these layer three links that required 108 05:07.850 --> 05:09.640 layer two connectivity. 109 05:09.640 --> 05:13.440 So we had to go in and say, well, okay, this was a good design. 110 05:13.440 --> 05:15.680 When it was made, it was layer three at the time. 111 05:15.680 --> 05:19.520 But now we're going to have to convert this to layer two, which was a headache 112 05:19.520 --> 05:20.080 that wouldn't 113 05:20.080 --> 05:23.560 have been needed if we just deployed layer two to begin with. 114 05:23.560 --> 05:25.920 And so that's something to consider as well. 115 05:25.920 --> 05:34.480 It is absolutely more complex as well from a routing protocol perspective. WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 115 05:25.920 --> 05:34.480 It is absolutely more complex as well from a routing protocol perspective. 116 05:34.480 --> 05:39.010 Remember my distribution layer was running maybe a stub version of a routing 117 05:39.010 --> 05:39.760 protocol, 118 05:39.760 --> 05:41.400 but it can't do that anymore. 119 05:41.400 --> 05:45.800 I'm going to have to deploy OSPF or EIGRP everywhere in this environment. 120 05:45.800 --> 05:50.830 I might have to deal with OSPF areas, make it so that each one of my closets is 121 05:50.830 --> 05:51.360 a stub 122 05:51.360 --> 05:52.360 area. 123 05:52.360 --> 05:57.010 It's going to be more complexity here, not necessarily complexity we can't 124 05:57.010 --> 05:58.040 handle, but it is 125 05:58.040 --> 06:02.820 going to be more complex and therefore it's going to be trickier to support WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 125 05:58.040 --> 06:02.820 going to be more complex and therefore it's going to be trickier to support 126 06:02.820 --> 06:03.480 issues that 127 06:03.480 --> 06:04.480 arise. 128 06:04.480 --> 06:11.120 And lastly, we do need to consider once again that wireless roaming concept. 129 06:11.120 --> 06:15.410 Even though we have a lot of our wireless deployed today such that we're using 130 06:15.410 --> 06:16.440 controllers, 131 06:16.440 --> 06:21.310 if I don't have a wireless LAN controller type of architecture, whether it says 132 06:21.310 --> 06:21.920 go or 133 06:21.920 --> 06:27.220 otherwise, this is going to be a major downside because if I have an access 134 06:27.220 --> 06:28.400 point here and an 135 06:28.400 --> 06:33.120 access point over here and we know that we're trying to create one large WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 135 06:28.400 --> 06:33.120 access point over here and we know that we're trying to create one large 136 06:33.120 --> 06:34.560 wireless network, 137 06:34.560 --> 06:38.940 well now when I roam from one closet to the other, this is going to be a layer 138 06:38.940 --> 06:39.400 three 139 06:39.400 --> 06:41.000 roam. 140 06:41.000 --> 06:45.060 Now again, if we have controllers that play, we can absorb that and that won't 141 06:45.060 --> 06:46.640 be a problem, 142 06:46.640 --> 06:51.680 but if it is a layer three roam, then that's going to cause us a lot of issues. 143 06:51.680 --> 06:57.280 So at some point we do have to stop and ask, okay, well which one is best then? 144 06:57.280 --> 07:03.240 And as you might imagine, the answer is going to be, well, it depends. WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 144 06:57.280 --> 07:03.240 And as you might imagine, the answer is going to be, well, it depends. 145 07:03.240 --> 07:07.530 We have a list of pros and cons for both layer two and layer three and we're 146 07:07.530 --> 07:08.120 going to have 147 07:08.120 --> 07:12.120 to get together with our team, with our potentially with our managers and 148 07:12.120 --> 07:13.200 discuss the pros and 149 07:13.200 --> 07:18.440 cons of each solution to make sure that we have the correct one deployed. 150 07:18.440 --> 07:24.580 I will say that the layer two deployment is usually considered to be the safest 151 07:24.580 --> 07:25.280 because 152 07:25.280 --> 07:27.080 it's nice and cheap. 153 07:27.080 --> 07:29.960 We all know how layer two works to the axis layer. 154 07:29.960 --> 07:33.930 We've been doing it for a long time and it's going to again maximize that WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 154 07:29.960 --> 07:33.930 We've been doing it for a long time and it's going to again maximize that 155 07:33.930 --> 07:34.760 flexibility 156 07:34.760 --> 07:39.190 so we don't have to worry about things changing in the future and having to re 157 07:39.190 --> 07:40.000 vert back to 158 07:40.000 --> 07:42.560 layer two if we were to go with layer three. 159 07:42.560 --> 07:47.010 But the fact that layer three gets rid of spanning tree, it's absolutely going 160 07:47.010 --> 07:48.120 to create 161 07:48.120 --> 07:53.280 the cleanest environment and therefore it's going to improve supportability 162 07:53.280 --> 07:53.600 from the 163 07:53.600 --> 07:57.960 perspective that we should have fewer issues or at least we might expect that. 164 07:57.960 --> 08:02.190 And so these are broadly speaking why we might want to go with one over the WEBVTT X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000 164 07:57.960 --> 08:02.190 And so these are broadly speaking why we might want to go with one over the 165 08:02.190 --> 08:02.320 other but 166 08:02.320 --> 08:06.740 at the same time again we have to compare the pros and cons of each solution 167 08:06.740 --> 08:07.400 and make the 168 08:07.400 --> 08:10.280 decision that is best for our organization.