CCNP Lab, Part 3 – Which L2 Switch?

LAN switches could be the largest expense in a CCNP lab. In keeping with the goal of looking at CCNP labs with the cost being one of the most important factors, today I'll look at three layer 2 switches - the 2900XL series, the 2950 w/ standard image, and the 2950 with enhanced image - all with an eye towards doing the functions covered for the BCMSN exam.

I settled on comparing these three options in part because that's where the CCNA discussions led us in our efforts to keep the cost low while still getting something out of the hardware. Briefly, the 2900XLs are cheap, and have many of the functions needed for CCNA level switching, but the syntax of the commands often differ from the 2950 and 2960 series L2 switches. 2950's have 2 software loads, standard and enhanced, as determined by the model of switch - you can't field upgrade to use the enhanced features. 2960s would be even better, but they are price-prohibitive compared to our goal of keeping the cost down.

For perspective, my earlier research from last month put 2900 XL's at around $100, 2950 standard at around $200, and 2950 enhanced at around $300.

So, to pick a switch or two, the trick is to figure out what you get relative to the CCNP exams with each of the switches. So, I started by looking at this URL, which is a Cisco document that compares the 2900XLs to 2950's with both images. I then compared the feature list to the BCMSN exam topics, and then looked for features that were supported on only a subset of the 3 types of switches. I ignored the features that weren't on the BCMSN list, and I ignored features that were supported on all 3 or not supported on all 3. Here's what I came up with; I may have missed a few.

First, consider the two cheaper switch options. The table highlights in blue the items you don't get on the 2900XL, but do get (or get more of for QoS) on the 2950 standard image switches. It's not a long list, but the Spanning Tree Protocol (STP) topics, plus the relatively small amount of QoS functions, makes the differences notable.

For further perspective, let me compare the first four features - all related to Spanning Tree Protocol (STP), versus the last 3 features. The STP features require multiple switches that support the feature (at least 2) in order to do any useful lab experiments. The rest of the features could be tested with a single switch capable of those features.

Next, let me throw out a bit of a curve, since it's getting towards baseball season - I found a few possible errors in the document that compares the 2900XL to the two 2950 images. As it turns out, in my home lab I have a WS-C2950T-24, which is an enhanced image 2950, and a WS-C2950-24, a standard image 2950. I could've sworn that both supported RSTP (aka 802.1w), which conflicts with the doc referenced above (and the table I listed above based on that document.) So, I did some quick tests, and indeed, I configured RSTP and MISTP on both the standard and enhanced image switches, no problems. So, let me state it again to be clear - from my testing, I think that the standard image 2950's support RSTP and MISTP just as much as the enhanced image.

Continuing the comparisons between the two images on the 2950's, the only other feature from my table that differs between the two 2950 images is QoS - and I know the enhanced image does indeed support many more QoS features. But from my experimentation, that's the only difference I can see between the two images that matters to the BCMSN exam.

So, summarizing what I've rambled about today:

1) 2900XLs often have different syntax in the commands to accomplish the same features, as compared to 2950/2960

2) 2900XLs are missing some features compared to the needs of the BCMSN exam

3) 2950 Std image switches are missing a few features compared to the BCMSN, but not as many as I first thought.

4) The difference between the 2950 images, at least for the features required for BCMSN preparation, seems to be pretty small.

So, anyone out there have a 2950 with standard image, and you're willing to test to confirm what I found? What am I leaving out? Any other perspectives out there on the L2 switches for CCNP?