No Layer 2 neither in TS or FS Lab

I had a look over the new released Labs, and I saw (I need to say I wasn't the first one to realize about this, but a former post about this in some other thread was surprisingly removed by the author), that there are no L2 tasks.

Even though you shouldn't expect all the technologies being present on all labs, L2 is important enough to be present in all of them. First, this is a Routing and Switching exam, so L2 and L3 not only are in the scope, but both of them are the "alma mater" of the track. And that's as simple as consider that one of the main lessons to learn by the CCIE candidate is that lower layer issues produce higher layer issues. Your BGP won't work well if your L3 (IGPs) isn't working well. And IGPs won't work well if your L2 isn't properly deployed and free of issues.

That's why we get TS questions like "A is not able to speak with B". Then we need to go to an structured approach for L2, L3 or even L4 troubleshooting and verifications till we get the root cause and fix the issue.

The question of concern for me is not getting L2 or not in these two first labs. The question of concern for me is that I realized after some other forum member disclosing this missing part that in my opinion it's going to be more difficult to get challenging L2 scenarios in the FS and TS labs with the physical topology INE is proposing for the v5 blueprint.

Why? Well, in this topology all routers are connected to a virtual hub in the Hypervisor, then connecting this Hypervisor "in a stick" to the L2 switches via a unique interface. So you don't really need the L2 switches for them to have reachabiliy to each other. In other words, traffic between routers does not traverse the switches segment, as it remains local in the hypervisor. Traffic that would be traversing the switches segment is limited to that going to/from SVIs in the switches. So any task or issue in the switches is not affecting them. This constrains the possibilities in TS, as it gets harder to trigger for example OSPF issues being triggered by L2 disruptions. We could technically do it between SVIs, but we know that we couldn't get more complicated scenarios...

Concluding, for example, we can almost 100% of the times discard any L2 issue in the TS tickets to begin with, which I don't believe that is the scenario I will face in the exam, and no matter my mistakes in L2 configurations they won't break anything in higher layers in my CFG lab, which I'm absolutely sure it's not going to happen that way in the exam.

My suggestion to INE would be to use at least one different physical link from the Hypervisor to each one of the switches so you can at least simulate them being connected to different physical portions of the network. That would lead us to more realistic scenarios, and proposed labs.

Opinions?

Comments

  • This is one of those cases where I'm not sure there is a good solution. We already know that the exam is all virtual now and that it is supposedly being delivered using IOS on Linux. The layer 2 version of this is buggy - which is something that worries me about the real lab. I hope Cisco has a super-secret non-leaked version that actually functions. The one I've seen can't even handle a triangle topology without going to 100% CPU and dropping packets (even though STP shows proper convergence).

    If INE wants to be able to provide a topology that scales to the level of the real lab it could be a bit of a challenge to tie in switches and have to maintain so many physical devices. Getting rid of physical routers probably helps to some degree, but they've certainly had to add a lot of servers to run the virtual routers. I highly doubt there is a legal option for them running IOU and then you would have to deal with all the bugs.

    I have to think that layer 2 is going to be scaled back quite a bit in the lab, but I wouldn't treat it as if there won't be something going on there. Maybe they could keep the 4 switches from v4 option and just not tie those into the routers and have a L2 section that is independent of the L3. That's not a perfect solution but it's bound to be better than just ignoring L2 altogether.

  • This definitely does decrease complexity possible with TS/config and I agree that by running a cable from the hypervisor to each switch (each connected to 4 different vswitches and separate vnics), you could have a topology that allows more complexity. Adding in a single quadport NIC would be viable for most people already building servers.

    I've been doing a lot of screwing around trying to figure out what the best solution is, and frankly, I think it's to mix the v4 labs into your prep. If you don't want to mess around with FR, FRTS, L2QOS, etc and you're not already solid on those topics, just paste those parts of the configs in if needed for a functional setup or skip them altogether if not. There's only 1 of each v5 FL, TS, FS at this point, so we also don't have much choice in the matter. Once there are a decent number of v5 labs, figure out what the optimal mix is based on your strengths and weaknesses.

    It does suck that you have to keep two platforms running thuogh. I think most would prefer to have a single topology and the labs to go with it that handled everything well. I'm probably going to suck it up and buy a 3750 so that I can more easily switch between v4 and v5.

  • Not all labs will include layer 2 switching. Traditionally layer 2 switching has never been a large focus of the CCIE R&S track. 

    Brian McGahan, 4 x CCIE #8593 (R&S/SP/SC/DC), CCDE #2013::13
    [email protected]
     
    Internetwork Expert, Inc.
    http://www.INE.com

    On Oct 5, 2014, at 4:50 PM, "bengood24" <[email protected]> wrote:

    This is one of those cases where I'm not sure there is a good solution. We already know that the exam is all virtual now and that it is supposedly being delivered using IOS on Linux. The layer 2 version of this is buggy - which is something that worries me about the real lab. I hope Cisco has a super-secret non-leaked version that actually functions. The one I've seen can't even handle a triangle topology without going to 100% CPU and dropping packets (even though STP shows proper convergence).

    If INE wants to be able to provide a topology that scales to the level of the real lab it could be a bit of a challenge to tie in switches and have to maintain so many physical devices. Getting rid of physical routers probably helps to some degree, but they've certainly had to add a lot of servers to run the virtual routers. I highly doubt there is a legal option for them running IOU and then you would have to deal with all the bugs.

    I have to think that layer 2 is going to be scaled back quite a bit in the lab, but I wouldn't treat it as if there won't be something going on there. Maybe they could keep the 4 switches from v4 option and just not tie those into the routers and have a L2 section that is independent of the L3. That's not a perfect solution but it's bound to be better than just ignoring L2 altogether.




    INE - The Industry Leader in CCIE Preparation

    http://www.INE.com



    Subscription information may be found at:

    http://www.ieoc.com/forums/ForumSubscriptions.aspx
  • To clarify, what I mean is that layer 2 switching will be in both the TS and full scale labs, just not in all of them. 

    Brian McGahan, 4 x CCIE #8593 (R&S/SP/SC/DC), CCDE #2013::13
    [email protected]
     
    Internetwork Expert, Inc.
    http://www.INE.com

    On Oct 5, 2014, at 4:50 PM, "bengood24" <[email protected]> wrote:

    This is one of those cases where I'm not sure there is a good solution. We already know that the exam is all virtual now and that it is supposedly being delivered using IOS on Linux. The layer 2 version of this is buggy - which is something that worries me about the real lab. I hope Cisco has a super-secret non-leaked version that actually functions. The one I've seen can't even handle a triangle topology without going to 100% CPU and dropping packets (even though STP shows proper convergence).

    If INE wants to be able to provide a topology that scales to the level of the real lab it could be a bit of a challenge to tie in switches and have to maintain so many physical devices. Getting rid of physical routers probably helps to some degree, but they've certainly had to add a lot of servers to run the virtual routers. I highly doubt there is a legal option for them running IOU and then you would have to deal with all the bugs.

    I have to think that layer 2 is going to be scaled back quite a bit in the lab, but I wouldn't treat it as if there won't be something going on there. Maybe they could keep the 4 switches from v4 option and just not tie those into the routers and have a L2 section that is independent of the L3. That's not a perfect solution but it's bound to be better than just ignoring L2 altogether.




    INE - The Industry Leader in CCIE Preparation

    http://www.INE.com



    Subscription information may be found at:

    http://www.ieoc.com/forums/ForumSubscriptions.aspx
Sign In or Register to comment.