Completion of the Analysis of a 2-Switch Lab for CCNP
The big question we’ve been looking at for the last few weeks is the value of a 2 switch lab topology for prepping Spanning Tree topics for both CCNA and CCNP. I’ll conclude the discussion this week, and ask for your help to build a list of “but you can’t do ______ with a 2-switch lab”. In short, though, I think you can practice most STP-related config for CCNP SWITCH, with the exception of 802.1s MSTP. The STP topology may be a bit boring, but you can practice the commands.
To get a sense for some of the specifics, you have to refer to the following topology, and to the lab details covered in the last two posts (here and here).
The lab described in these previous posts boils down to this:
- Cause VLAN 2’s STP topology to send traffic over the top trunk
- Cause VLAN 3’s STP topology to send traffic over the bottom trunk
- Do so with as many different configuration combinations as possible
If you ignore the defaults of all the commands, the configuration options in both VLAN 2 and VLAN 3 reduce to the same issues and same possible configurations, other than those taking advantage of default settings. Since VLAN 3 required the fewest default settings, I’ll focus there today.
First, I need to answer the 5 questions I posed last blog post – answers that tell us more about which features could be configured to impact the STP topology:
1) The STP cost of a portchannel (etherchannel) defaults based on the number of currently working links. As such, you could choose to experiment by adding links, and shutting them down.
2) In VLAN 2, by default, SW1 becomes root, so SW2 chooses either its portchannel (upper) or FastE (lower) trunk as its root port. If the cost of SW2’s Portchannel and its F0/10 were set to tie, the tiebreaker is the port priority in that VLAN. With default settings, both the portchannel and F0/10 tie (128). The last tiebreaker is the internal port number. Portchannels use (by default) higher numbers than the physical ports on a switch, so in this case, SW2 would choose its F0/10 interface as its root port.
3) The short answer is “no”. Once SW1 has been elected root, it will advertise Hellos, cost 0, to SW2, over both trunks. Even though you could configure the costs differently on SW1 on the trunk ports, it does not affect the cost listed in the Hellos sent to SW2, so the configuration on SW1 does not affect SW2’s choice of Root Port.
4) A change in link speed changes the default STP cost, even when a link is in an active portchannel. So, on trunks where the cost is not explicitly set, setting the speed impacts the cost. On port channels, all link speeds must be the same, so it also impacts the number of links in the port channel, again impacting STP cost. For example, if both links in the portchannel were changed from 100 Mbps to 10 Mbps, the portchannel changes from cost 12 to cost 56.
5) Enabled Multiple Spanning Tree Protocol (MST, aka MIST and MSTP), 802.1s, opens up just as many options for configuration as the traditional 802.1d config discussed so far in this series. You could configure both switches in the same MST region, and essentially have just as many config options as with 802.1D, just with slightly different syntax. You could also, just for the same of experimentation, configure 1 switch for MST, and the other not, but frankly it’s pretty dull, but it can be configured.
In short, the 2-switch topo gives you lots of configuration options. The STP topology is much more interesting with 3 or 4 switches, and the show commands can give you a little more interesting output.
After writing all this, I looked at the SWITCH exam topics, and looked at Dave’s CCNP SWITCH Official Cert Guide. The exam topics, as usual, don’t reveal enough detail to matter. In fact, “Spanning Tree” isn’t even mentioned, although it does mention “layer 2 loop prevention”. However, roughly ¼ of the SWITCH book examines STP-related topics. So I scoured Dave’s book, and looked for topics that couldn’t be meaningfully configured at all in a 2-switch topo. IE, MSTP config is pretty dull with two switches, but you can indeed configure and test it with 2 switches. I found only two STP features in the loop that couldn’t be configured and tested meaningfully on this 2-switch topo:
- Backbone Fast
- Uplink Fast
Both require an indirect link to the root, which by definition requires a 3rd switch.
In summary, I think you only need two switches for STP practice for CCNA and CCNP. A 3rd switch, or even a 4th, makes the exercises a lot more interesting, but you don’t need the 3rd switch for a lot. So – what am I leaving out? What other STP-related stuff would you not be able to do, or feel much constrained in doing, with just 2 switches?
