My thoughts on the network evolution

Recently, I’ve been putting a lot of thought into where the ‘next big thing’ in networking is going to be.  Now, I’m not just pondering these thoughts for no apparent reason.  I was on the fence between focusing on new technology and getting my CCIE Route Switch.  The CCIE has been a dream of mine for quite some time and as such, it was hard to think about giving it up to focus on new networking technology.  On the flip side, getting involved with some of these new ideas, concepts, and tech is hard to pass up.  Anyone with enough ambition has the chance to get in on the ground floor.  And don’t get me wrong, there has been some really interesting technology that’s been coming out in the last year with no real slow down in sight.  So I was at a bit of a cross road. 

Luckily for me, I think I’ll be able to balance myself between the two.  I’m fortunate enough to have a job that includes new networking technology evaluation as part of the role.  This leaves what free time I have to pursue my CCIE during non-working hours.  However, while I was on the fence about putting all of my emphasis one way or the other, I came across an interesting idea.  There are lots of people who have one (or more) of the following beliefs..

The CCIE certs (particularly R&S) will be a thing of the past.  Engineers with CCIE certifications will no longer be in demand. 

New technology coming out will all but replace archaic CLI driven network devices.

The cloud is going to take over the world.  If your stuff doesn’t run in the cloud, it will in the next year. 

If you a network engineer, your time is better spent starting in on the DevOps movement and learning programming languages like Python. 

I’m not quoting anyone here, but I think you get my drift.  People are going SDN crazy and think the world of networking is about to do a complete 360.  I disagree.  In fact, I disagree entirely will all of the beliefs I listed above.  Let me walk through them one at a time and give my reasoning…

The CCIE certs (particularly R&S) will be a thing of the past. Engineers with CCIE certifications will no longer be in demand.
I’ve thought about this one for a particularly long time.  Every time my thought process starts pushing me in the ‘CCIEs are going away’ direction all I have to do is to go back to work.  I work for a large company and I use the material covered in the CCIE R&S series of tests every single day.  BGP, OSPF, Spanning-Tree, VLANs, MPLS, etc are NOT going to go away.  In fact, as we add more infrastructure the need for engineers with these skill sets are going to increase.  If you are on the cloud or massively distributed data center kick you are going to NEED people who understand the right way to make data centers scale.  I could give more examples of this, but I think you get the point. 

New technology coming out will all but replace archaic CLI driven network devices.
Anyone that thinks that large providers or enterprises are going to ‘rip and replace’ all their existing (functioning) infrastructure are verging on the edge of madness.  Not only would the cost be astronomical, it just wouldn’t make sense.  I haven’t seen one piece of new networking tech that I believe would justify me ripping out my WAN head-ends, core switches, and remote WAN routers.  My point is that some of this new technology (OpenFlow comes to mind here) might help me get better ‘network’ utilization by enhancing some of the control plane activities.  However, the router’s will still be the same and they will still be managed the same way.  If you think that these networking devices are going to be managed by some central piece of software I’d disagree with that too.  I mean, look at Cisco’s attempts to do that this far (CSM for ASAs comes to mind).  Did that work out so well?  What happens when there’s a problem with the firewall?  You still have to log into the CLI to figure it out.  I also want to bring up an experience last year when I had the opportunity to tour a large carrier’s POP.  The amount of legacy infrastructure in their building took me by complete surprise.  Rooms full of older TDM equipment that they still had to maintain and still had to have people who could manage it.  When I asked about it they didn’t see it going away any time soon.  T1 and other TDM circuits still get deployed and will continue to be. 

The cloud is going to take over the world. If your stuff doesn’t run in the cloud, it will in the next year.
While I see the use cases for ‘cloud-like’ services, I don’t foresee any large enterprise moving all of their critical Tier-1 apps into a public or private cloud.  In fact, many people won’t even move their critical apps to any sort of virtualized infrastructure.  While the cloud has it’s advantages, I see it more as a smaller use case, quick spin up of development VMs, sort of thing.  Vendors that come in and tell me how ‘things are going to be’ and that I ‘need their tech’ to keep doing business next year are lacking the big picture view.  The world of networking will keep spinning with or without the cloud. 

If you a network engineer, your time is better spent starting in on the DevOps movement and learning programming languages like Python.
Let me be clear about this one.  I am NOT against network engineers broadening their skill set.  I recall when I learned Perl and PHP for the first time three years ago when I started my new job.  One of my coworkers was using to for all kinds of really slick configuration automation.  It was really, really, really cool!  I quickly caught on and had quite a bit of success using it to make my job easier.  My problem with this statement is the ‘better spent’ piece.  From my viewpoint, as a network engineer, your job first and foremost is to be good at the network element.  That being said, my belief is that your time ,as a network engineer, is best spent being really solid at configuring and troubleshooting the network.  This includes evaluating new technology and understanding how this new technology can impact other silos (storage, compute, security, etc).  In the end, it’s your job to be the network expert.  If you can learn other stuff along the way (python, perl, etc) that’s awesome, but don’t put that stuff in front of being good at your specific job.  If you’ve mastered all that boring network stuff, then spread your wings and go for it. I should clarify here that I’m talking about a network engineer that’s just a network engineer.  If you work somewhere where you are ‘IT’ then this obviously doesn’t work. 

All of that said, we are living in a very exciting time in regards to networking!  Things are changing and I believe they will continue to change to make our jobs more and more interesting.  Some of this new tech will make it through the concept phase and see deployment in large enterprises.  Some of it already has and some of it will likely die before it sees it’s first large customer deployment. 

I don’t want to come across as a luddite, what I want to do is be realistic.  I also want to drive the point that there are lots of ways to skin the proverbial cat with networking.  Take for instance the case for layer2 multi-pathing (Fabric Path for example).  I recall talking to an engineer (who worked for a particular vendor *Cough* Cisco *Cough*) at a conference when Fabric Path has just been announced.  I was talking about how I didn’t see the use case for it.  This engineer just didn’t understand how I didn’t see the technology as useful.  After I explained to him that we ran layer 3 to the edge, explained how we don’t have spanning tree blocking uplinks, and that our spanning-tree domain was the two edge switches I could see him thinking about it and he finally just said  “Huh…”.  My point is that we all have different networks.  Different technology makes sense to deploy in your network and doesn’t make any sense at all to deploy in mine. 

So in summary I wanted to make a few points…

-‘Normal’ networking (networking as I know it today) is not going to go away any time soon.  New technology will come out and be deployed, but it will likely ride on top of existing network infrastructure or compliment it. 

-The CCIE (and all other certs) will continue to maintain it’s value.  Even if you replace the whole DC network with something new and shiney, traditional networking will still exist (WAN and provider networks are great examples).  As long as that tech exists, we’ll need people to deploy, configure, and troubleshoot it.  If people start jumping off this bandwagon as some people imply, the demand will be even higher. 

-It’s your job as a network engineer to look at all of this new technology and see what makes sense in your environment.  Can this new tech add new value?  If so, in what piece of the network? 

-Virtualization has certainly been a game changer.  It makes sense for networking engineers to start being more interested in virtualization as the networking component becomes more defined.  Just make sure you aren’t skipping reading that spanning-tree chapter of your CCNA book and jumping right into stretching layer 2 across your data centers.  Make sure you know the basics before you jump into the cool new stuff.

So there it is.  Those are my thoughts.  I’d love to hear other peoples input on this.  And who knows, maybe I’ll be proven completely wrong!

11 thoughts on “My thoughts on the network evolution

  1. that1guy15

    “In the end, it’s your job to be the network expert. If you can learn other stuff along the way (python, perl, etc) that’s awesome, but don’t put that stuff in front of being good at your specific job. ”

    Nail right on the head.

    Great post!!

    Reply
  2. Will Dennis

    I just wish that there were some advanced non-vendor-specific networking courseware / testing out there, instead of just the Cisco-centric CCNP / CCIE stuff (yes I know other vendors like Juniper, HP, etc have theirs.) I totally agree one needs to know the basics of the protocols, design implementation choices, etc. for routing and switching, but I just wish there was some courseware disassociated from the vendor-specific offerings out there. I’m very excited about what Cumulus Networks just announced, and OpenFlow and SDN in general, which may take us from the vendor lock-in we experience today, to a more open future where we may use something such as Linux, Open vSwitch, OpenFlow, etc. on the hardware of our choice (at a much better price point than we currently experience!) to provide the business with needed network services. Perhaps someday something like OpsSchool ( https://ops-school.readthedocs.org/en/latest/ ) can fill the need for this vendor-agnostic training.

    Also, I myself am trying to straddle the divide between traditional systems admin (i.e. server infrastructure and installing/ configuring software on top of it) and network administration. I am studying for a Cisco exam, learning Python, and (soon) taking a course on SDN all at the same time… All while working at my day job as a systems/network admin. I do think there will always be a place for the deep-but-narrow CCIE-type experts, but am betting that increasingly there’ll be more of a market for cross-discipline folks who understand both server *and* network admin, and know how to program at least in a scripting language such as Python, Ruby, etc. since virtualization (both server and network) and cloud technologies will demand this. I guess it all depends on what you want to focus on at the end if the day…

    Reply
    1. Jon Langemak Post author

      I agree with you on all counts. At the end of the day, it depends on what kind of environment you are expected to build and support. In my case, I dont see an end to the need for CCIE R&S type experts. That’s technology that I dont think will ever go away in large enterprise networks. Will new technology come that might supplement or slightly change how those networks are built? Likely, but I don’t see the end of that type of engineering any time soon. The networking world is at such a volatile state right now it’s hard to bet the farm (your career path) on any one new technology. In the end, I think we can all agree that we are living in VERY interesting times as far as the network is concerned. It will be interesting to look back in 10 years and see what tech is still around and what tech died on the vine. Exciting!

      Reply
  3. Daniel Dib

    Agree on all parts. There will still be need for people that understand how things really work.

    Also agree on the programming. Yes, I would like to learn some programming but I want to be as good an engineer as possible first. I’ll leave coding to people that are good at it.

    Reply
  4. Josh Barron

    I’m starting a coursera course on SDN Monday evening. I’m looking forward to it to broaden some of my networking skills as I continually end up doing more systems engineering (virtualization, etc) in my work in consulting.

    I’ve had numerous conversations with other network engineers in my comings and goings. Some have never even heard the term “SDN”, and others think it will end world hunger and solve the crisis in Syria.

    Like you, I’m on the fence. I look forward to learning things that will help move things forward, while realizing at the end of the day the core concepts will remain. Programmers will never be expert network engineers, nor will we ever really be expert programmers (at least at the same time that is).

    Reply
    1. Jon Langemak Post author

      Im actually taking that course too! I had signed up for it so long ago that I had forgotten (and I didnt get a reminder email (weird)).

      Yeah, I think it’s fair to say that Im skeptical about how much will actually change in the next few years. Im certain some things will change for the better, but I dont think anyone is going to totally change the way we build the majority of our networks in the next 2 years.

      Reply
  5. windexh8er

    Here’s the rub; great network engineers will always be in demand – that’s a given. The construct of when and where they are in demand is the outlier. 10 years ago, CCIE guaranteed you a slot anywhere in a big environment. Today, that’s unfortunately, not true.

    The critical path is being able to balance what you’re getting from the CCIE studies and how you apply it to being agile in the workforce going forward. Could I go work on a straight network team today? Yes – I can hold my own in any large BGP/OSPF/MPLS heavy environment – the platforms are not the problem (classic IOS/IOS-XR/etc.) it’s applying the foundations back into the configuration. However, I’ve run across many non-Cisco platforms more in the past 3 years than ever before. Can I configure BGP on JunOS? Yes – partly because the foundational knowledge I had from my Cisco studies affords me this luxury.

    I know you are trying to balance out your path forward – but I’ve bumped, first hand, into the limitations of playing in the vendor-specific sandbox. Why is it great to know languages that play well into the “SDN” realm? Not because any one language is a requirement, but if you understand how to leverage a language to make an API call to provision/change/influence the network-as-a-whole you’re ahead of 80% of your peers. Will all of those new-fangled technologies work out? No, and hell-no at that. But, expanding yourself into a position that’s technology and vendor-agnostic has it’s wins. Most startups stay away from Cisco – why? Because they’re expensive and, one can argue today, that there’s not as much positive outcome from using them. I’ve seen, first hand, having Cisco experience on my resume actually hinder potential opportunities because of the way those smaller companies approach keeping arms length from people who are pigeon-holed in thinking and approach to new or out-of-the-box technology.

    TL;DR
    Don’t bet your career path on a vendor technology, but take from that vendor the core components that will lend progress to your career long-term. If you’re a network phenom who can write in an object oriented language to further make themselves more viable/efficient – you’re resume is going to go to the top of the 80% of the people who are fine with playing in a very defined sandbox. And, finally, you don’t end up a network rockstar staying in one environment unfortunately… You also, don’t climb the fiscal ladder quickly either. 😉

    Reply
    1. Jon Langemak Post author

      An interesting take with valid points. When I started networking I spent some a lot of time looking around trying to figure out ‘how to’ learn to be a network engineer. After a lot of poking around, I finally just decided to get my CCNA. What I found surprised me. The CCNA books (at least back then) did a VERY good job teaching you what a network was and THEN teaching you how to configure those networks on Cisco gear. Is that the case any more? Likely not. Cisco’s cranking out more platforms every year than ever before (which is really really really annoying). Learning one platform (IOS) doesn’t guarantee you anything so I agree with you there. However, if you are smart enough to know that you have to abstract the technology (BGP, MPLS, OSPF, etc) from the equipment (Cisco) you can still learn quite a bit. This could easily go into the ‘standards are important discussion’ as well but that’s for another post. I’d love to work on other vendors tech but unfortunately I only get Cisco at work to play with.

      The bottom line is that ,like you said, you need to make a choice about where you want to go with your career. At this stage of the game betting all of your money on one horse is a mistake in my opinion but we’ll just have to wait and see if Im right or wrong.

      Great conversation by the way.

      Reply
  6. Joe Brunner

    The the logic, speed and skill of the CCIE make it a worthwhile investment EVEN if you NEVER do cisco after the day you pass it 🙂

    it re-wires your brain to be the IT version of Tom Cruise in the club scene in Collateral.

    🙂

    Trust me 🙂

    Reply

Leave a Reply to Joe Brunner Cancel reply

Your email address will not be published. Required fields are marked *